MIR - Faddeev D. K. and Sominsky I. - Problems in Higher Algebra - 1972

A. K. (DALLEIEEB, H. C.
COMPIHCKHrl
CBOPHMK
no
BBICLUEn AJ1FEBPE
1134ATEAbCTBO HAYKA MOCKBA
D. FADDEEV, I. SOMINSKY
PROBLEMS IN HIGHER ALGEBRA

TRANSLATED FROM THE RUSSIAN by GEORGE YANKOVSKY
MIR PUBLISHERS MOSCOW
UDC 512.8 (075.8)=20
Revised from the 1968 Russian edition
Ha atteiuttiocom
ii3b1Ke
TO THE READER Mir Publishers would be grateful for your comments on the content, translation and design of this book. We would also be pleased to receive any other suggestions you may wish to make. Our address is: Mir Publishers, 2 Pervy Rizhsky Pereulok, Moscow, USSR.
Printed in the Union of Soviet Socialist Republics
Contents Introduction Part I PROBLEMS

CHAPTER I. COMPLEX NUMBERS
11 11 13 19 21 25 25 26 27 29 31 51 56 61 61 64 66 68 76 76 83
1. Operations on Complex Numbers 2. Complex Numbers in Trigonometric Form 3. Equations of Third and Fourth Degree 4. Roots of Unity
CHAPTER 2. EVALUATION OF DETERMINANTS
1. Determinants of Second and Third Order 2. Permutations 3. Definition of a Determinant 4. Basic Properties of Determinants 5. Computing Determinants 6. Multiplication of Determinants 7. Miscellaneous Problems
CHAPTER 3. SYSTEMS OF LINEAR EQUATIONS
1. Cramer's Theorem 2. Rank of a Matrix 3. Systems of Linear Forms 4. Systems of Linear Equations
CHAPTER 4. MATRICES
I. Operations on Square Matrices 2. Rectangular Matrices. Some Inequalities

CHAPTER 5. POLYNOMIALS AND RATIONAL FUNCTIONS OF ONE VARIABLE
88
1. Operations on Polynomials. Taylor's Formula. Multiple Roots 88 2. Proof of the Fundamental Theorem of Higher Algebra and Allied Questions 92 3. Factorization into Linear Factors. Factorization into Irreducible Factors in the Field of Reals. Relationships Between Coefficients and Roots 93
6 4. Euclid's Algorithm 97 5. The Interpolation Problem and Fractional Rational Functions 100 6. Rational Roots of Polynomials. Reducibility and Irreducibility over the Field of Rationals 103 7. Bounds of the Roots of a Polynomial 107 8. Sturm's Theorem 108 111 9. Theorems on the Distribution of Roots of a Polynomial 10. Approximating Roots of a Polynomial 115
CHAPTER 6. SYMMETRIC FUNCTIONS 116
I. Expressing Symmetric Functions in Terms of Elementary Symmetric Functions. Computing Symmetric Functions of the Roots 116 of an Algebraic Equation 2. Power Sums 121 123 3. Transformation of Equations 124 4. Resultant and Discriminant 5. The Tschirnhausen Transformation and Rationalization of 129 the Denominator 6. Polynomials that Remain Unchanged under Even Permutations of the Variables. Polynomials that Remain Unchanged under Cir130 cular Permutations of the Variables
CHAPTER 7. LINEAR ALGEBRA
133
1. Subspaces and Linear Manifolds. Transformation of Coordinates 133 135 2. Elementary Geometry of n-Dimensional Euclidean Space . . 139 3. Eigenvalues and Eigenvectors of a Matrix 141 4. Quadratic Forms and Symmetric Matrices 146 5. Linear Transformations. Jordan Canonical Form
PART II HINTS TO SOLUTIONS

CHAPTER I. COMPLEX NUMBERS CHAPTER 2. EVALUATION OF DETERMINANTS CHAPTER 4. MATRICES CHAPTER 5. POLYNOMIALS AND RATIONAL FUNCTIONS OF ONE VARIABLE CHAPTER 6. SYMMETRIC FUNCTIONS CHAPTER 7. LINEAR ALGEBRA
151 153 159 160 164 166
7 PART III ANSWERS AND SOLUTIONS

CHAPTER I. COMPLEX NUMBERS CHAPTER 2. EVALUATION OF DETERMINANTS CHAPTER 3. SYSTEMS OF LINEAR EQUATIONS CHAPTER 4. MATRICES CHAPTER 5. POLYNOMIALS AND RATIONAL FUNCTIONS OF ONE VARIABLE CHAPTER 6. SYMMETRIC FUNCTIONS CHAPTER 7. LINEAR ALGEBRA INDEX
168 186 196 203 221 261 286 313
INTRODUCTION
This book of problems in higher algebra grew out of a course of instruction at the Leningrad State University and the Herzen Pedagogical Institute. It is designed for students of universities and teacher's colleges as a problem book in higher algebra. The problems included here are of two radically different types. On the one hand, there are a large number of numerical examples aimed at developing computational skills and illustrating the basic propositions of the theory. The authors believe that the number of problems is sufficient to cover work in class, at home and for tests. On the other hand, there are a rather large numb:x of problems of medium difficulty and many which will demand all the initiative and ingenuity of the student. Many of the problems of this category are accompanied by hints and suggestions to be found in Part I I. These problems are starred. Answers are given to all problems, some of the problems are supplied with detailed solutions.
The authors
PART I. PROBLEMS
CHAPTER 1 COMPLEX NUMBERS
Sec. 1. Operations on Complex Numbers

1. (1 +2i)x+ (3 5i)y = 1 3i. Find x and y, taking them to be real. 2. Solve the following system of equations; x, y, z, t are real: (1 +i)x+(1 +20y+(1 +3i) z+(1 +4i)t=1 +5i, (3 i)x + (4 2i)y + (1 + i)z + 4it =2 i.
3. Evaluate in, where n is an integer. 4. Verify the identity x4 +4=(x-1-0(xl+i)(x+1+i)(x+1-0.

5. Evaluate:
(a) (1 +2i)6, (b) (2 + 07+(2 07, (c) (1 +2i)5 (1 205. 6. Determine under what conditions the product of two complex numbers is a pure imaginary. 7. Perform the indicated operations: (b) a+ bi (c) ((3 +i tan a 1 ++ 2 )) 1 + )):
/ 1
i tan a
a bi
1 2
(d) (1 (1+05 +1 ' (e) (1 + 0n y 8. Evaluate (1 i
(1+ 09 (1 - 07
where n is a positive integer.
9. Solve the following systems of equations:
(a) (3 i)x+ (4 + 2i)y =2 +6i, (4 +2i)x (2 +30y-5+4i; (b) (2 + i)x+(2 i)y = 6, (3 + 2i)x + (3 2i)y = 8; (c) x+ yi 2z =10, x y +2iz =20, ix + 3iy (1 + i)z = 30.
12
PART I. PROBLEMS
10. Evaluate:
2
(a) ( (b)
2 1+ 1+
*11. Let co =
2 +
)3 . i 1,/ 3
2
. Evaluate:
(a) (a+bco+cw2)(a+b(o2+cco), (b) (a + b) (a + bco) (a +1,0), (c) (a + bo+ cco2)3+(a+ bco2+ c03, (d) (aco2+bco) (bco2+aco). 12. Find the conjugates of: (a) a square, (b) a cube. *13. Prove the following theorem: If as a result of a finite number of rational operations (i. e., addition, subtraction, etc.) on the numbers x1, x2,..., x, we get the number u, then the same operations on the conjugates .c x2,..., iT yield the number u, which is conjugate to u. 14. Prove that x2 + y2= (s2 t2 if x+ yi=(s + 15. Evaluate:
n )
(a) V 2i, (b) V 8i, (c) V 3 4i, (d) V 15 + 8i , (e) V 3 4i, (f) V 11 + 60i, (g) V 8 + 6i , (h) V 8 6i, (i) V 8 6i, (j) V 8 + 6i, (k) V 2 3i ,
4 / 1, (1) V4 + + V4 (m) V1 i V 3 ; (n) -1
(o) 4 1/2
16. V a + bi = (a + Pi). Find V a bi 17. Solve the following equations: (a) x2 (2+ i)x+ (-1 +7i)=0, (b) x2 (3 2i)x+ (5 5i)=0, (c) (2 + Ox2 (5 i)x+(2 2i)=0. *18. Solve the equations and factor the left-hand members into factors with real coefficients: (a) x4 + 6x3+ 9x2 + 100 = 0, (b) x4 +2x2 -24x +72 =O.
CH. 1. COMPLEX NUMBERS
13
19. Solve the equations: (a) x4 -3x2 + 4 =0, (b) x4 -30x2 +289 =O. 20. Develop a formula for solving the biquadratic equation
x4+px2 + =0 with real coefficients that is convenient for the case when - q < 0.
Sec. 2. Complex Numbers in Trigonometric Form 21. Construct points depicting the following complex numbers: 1, 1, 1/2,
i1/2, 1+i, 2-3i.
22. Represent the following numbers in trigonometric form: (a) 1, (b) 1, (c) i, (d) i (e) 1 +
,
(f) 1+ i, (g) 1 i, (h) 1 i (i) 1 + i V 3 ,

,
(j) 1 + il/ 3 , (k) 1 i 1/ 3 , (1) 1 i V 3, (m) 2i, (n) 3, (o) V 3 (p) 2+ V 3 + i. 23. Use tables to represent the following numbers in trigonometric form: (a) 3+ i, (b) 4 i, (c) 2+i, (d) 1 2i. 24. Find the loci of points depicting the complex numbers whose: (a) modulus is 1, (b) argument is
i Tt
25. Find the loci of points depicting the numbers z that satisfy the inequalities: (a) I zI <2, (b) zi 26. Solve the equations: (a) (c) z 1 I <1.
x x=1 +2i, (b)
+x=2+ i.
*27. Prove the identity
x +y12+1 x-y12 =
2(I x12 y 1 2) .
What geometrical meaning does it have?
14
PART I. PROBLEMS
*28. Prove that any complex number z different from 1, 1 + ti whose modulus is 1, can be represented in the form z= 1 u where t is real. 29. Under what conditions is the modulus of the sum of two complex numbers equal to the difference of the moduli of the summands? 30. Under what conditions is the modulus of the sum of two complex numbers equal to the sum of the moduli of the summands? *31. z and z' are two complex numbers, u=1/ zz'. Prove that = z+z'
2 u
, z+z'
I m1 2 1-
u
I
32. Demonstrate that if z
then
(1 + i)z3+ iz < 4 . 33. Prove that (1 +il/ 3) (1+0 (coscp+i sin co)= = 2 1/Y [cos ( 7 : 7+ cp) + i sin (-'7j + cp)] . 2 34. Simplify 35. Evaluate 36. Evaluate : (a) (1 + 025, irs (C) (I
2 cos cp + i sin cp cos (1) i sin (1, (1i V 3)(coscp+isincp) 2 (1 i)(cos cp i sin cp)
(b) (1 +1/ -v 3)2o, 1i \24

)
(d) (-1+ii/ 3)" (-1(1 _F i1 iro /
*37. Prove that
(a) (1 + = 2 (cos 4- + i sin 2'4`) 7 ,

(b) ( V 3 On =2" (cos 6 i sin 6 ,
Inc
n an integer.
CH. I. COMPLEX NUMBERS
15
27c
*38. Simplify (1 + w)n, where w =cos 3 + i sin 2, (0 2= 39. Assuming co, = 1 + i -1/S determine w7+ wz, where n is an integer. *40. Evaluate (1 +cos a+ i sin 0)n. *41. Prove that if z+ -I= 2 cos 0, then 1 zm + = 2 cos m0.
zm
-1/ 3 i2
42. Prove that
( 1 + itan oc )22 i tan a
1 + i tan na 1 i tan not
43. Extract the roots:

3 3 4 6 6
(a) V i
(b) 1/2 2i, (c) 1/ 4, (d)
1 , (e) V 2
44. Use tables to extract the following roots :

3 3 5
(a) 1/ 2 +
(b) V 3
(c) 1/ 2 + 3i.
45. Compute:
6 8 6
(a) / 1i V V3+i ,
(b)
+i s , (c)
1+i -V 3 .
46. Write all the values of V a if you know that B is one of the values. 47. Express the following in terms of cos x and sin x: (a) cos 5x, (b) cos 8x, (c) sin 6x, (d) sin 7x. 48. Express tan 6 cp in terms of tan cp. 49. Develop formulas expressing cos nx and sin nx in terms of cos x and sin x. 50. Represent the following in the form of a first-degree polynomial in the trigonometric functions of angles that are multiples of x: (a) sine x, (b) sin' x, (c) cos' x, (d) cos' x.
16
PART I. PROBLEMS
*51. Prove that m-1 (a) 22m cos2m x = 2
E ct cos 2 (m k) x +
k=0
(b) 22m cos2m +1 x=

k=0 in I
CL +1cos (2m 2k +1) X
(c) 2'm sin2m x = 2
k=0
z (_ om+k 0,ncos 2 (m k) x +
rn
(d) 2'm sin2m+1 x=E (_,),n+k Ct+Isin (2m 2k + 1) x.

k=0
*52. Prove that 2 cos mx = (2 cos x)m

1+2
(2 cos x)m -2 +
m m 3) 12
(
(2 cos X)m
m (m p 1) (m p 2) . . . (m-2p+1)
P!
x (2 cos x)in -2P

*53. Express sin x in terms of cos x. *54. Find the sums:
sin mx
(a) 1 0+0,-0+ ..., (b) ++...

n
*55. Prove that (a) 1 +0+0+...= 1 (b) 0, + (6) (d) + + 1 + 02 ,+.. = 2 + +...=
n -1
+2 2COS 17r 4)'

7- I , !--;),
(2n 1 2.7sin
2 0. cos 1-
'4' - . Cn " +... = 2 1 (2n -' 2 2 sin 17
17
*56. Find the sum

1 I ^
+ (x + aco)m + (x + aco2)m = 3 xm+ 57. Prove that (x + + i sin 27.` a' + . . . + 3 C'i, xnz-n an, where co = cos ' Xm-3 + 3 cn and n is the largest integral multiple of 3 not exceeding m. 58. Prove that
(a) 1 + + + . . . =
(2n 2
cos 7) ,
(b) Gin + C,1+

(c) +
+...=
1 1
(24+ 2 cos (n -3 ) (2" + 2 cos (n-3 )7

)
+ C,1+ . . . =
59. Compute the sums :
(a) 1 +a cos cp + a2cos 2y + +ak cos ky, (b) sin cp + a sin (y + h)+ a2 sin (so + 2h) + + ak sin (cp +kh), 1 (c) +cos x +cos 2x+ + cos nx. 60. Demonstrate that
sin
n+ 1 nx
sin x + sin 2 x + . . . + sin nx = 61. Find

n> oo
lira
2 x sin 2
sin 2
1 1 , 1 ( 1 + cos x + 71cos 2x + . .. + yn- cos nx) .
62. Prove that if n is a positive integer and 0 is an angle satisthen fying the condition sin = 2n,
cos 0 2n-1 30 + cos 2 + . . . + COS 2 0 = n sin ne.
63. Show that

5n 7 3n + cos 11 + cos i i = 2 , (a) cos -ff + cos 11 + cos 11 8n lOn 1 6n 2n 4n + cos 171 + cos -IT= T , i- i- + cos -171 (b) cos-ft+ cos -77
97
lln 1 7n 9n 5n 3n 2- . 3 -= + cos + cos --, + cos T (c) cos -T' + cos + cOs -13 13 16 13 13
18
PART I. PROBLEMS
64. Find the sums (a) cos a cos (a+ h)+ cos (a + 2h) +( On-1cos [a + (n
1)h],
(b) sin a sin (a + h)+ sin (a +2h) . . . + ( 1)"-' sin [a
+ (n 1)4
65. Prove that if x is less than unity in absolute value, then the series (a) cos a +x cos (a + (3)+x2cos (a + 2P) + +JO cos (a + nf3) + (b) sin a +x sin (a + co+x2 sin (a + 2P) + +xn sin (a + nP) + . converge and the sums are respectively equal to
cos a x cos(ap) 1-2x cos f3+ x2 ' sin a x sin (a--(3) 1 2x cos p +.x.
66. Find the sums of: (a) cos x + C';, cos 2x + . . . + C"t cos (n + 1) x, ni sin (n + 1) x (b) sin x + Oz sin 2x + . . . + e 67. Find the sums of: r; COS (n + 1) x , (a) cos x C';, cos 2x + C, cos 3x . . . + ( 1)n C (b) sin x 0, sin 2x + sin 3 N.. . . . + ( 1 )12 e' sin (n + 1) x
*68. 0A1 and OB are vectors depicting 1 and i respectively. From 0 drop a perpendicular 0A2on A1B; from A2 drop a perpendicular A2 43 on 0,41; from A3, a perpendicular A3A4 on A1 A2, etc. in accordance with the rule: from A a perpendicular Ani=1 + , is dropped on An _ zAn _1. Find the limit of the sum 0.41 + Av4 2 + A 2 A3 +...
*69. Find the sum sin2 x + sin2 3x + 70. Show that: (a) cos2 x + cos2 2x + ... + cos2nx = 2 (b) sin' x + sin2 2x + ... + sin2nx =
n
+ sin2 (2n 1)x.
cos
(n + 1) x sin nx
2 sin x cos (n + 1)x sin nx 2 sin x
19
*71. Find the sums of: (a) cos3x+ cos32x+ ... +cos' nx, (b) sin3x+ sin32x+ ... +sin" nx.
*72. Find the sums of:
(a) cos x +2 cos 2x +3 cos 3x+ ...+n cos nx, (b) sin x +2 sin 2x + 3 sin 3x+ +n sin nx. 73. Find lim (1 +
n> co )n
for oc = a+ bi.
74. Definition: e =lim (1 + n. Prove that (a) e2nr =1, (b) (c) ea +13 =e 0, = 1, (d) (e)k =ek for integral k.
Sec. 3. Equations of Third and Fourth Degree

75. Solve the following equations using Cardan's formula:
(a) x3 6x +9 =0, (b) x3+ 12x +63 =0, (c) x3 +9x2 +18x+28 =0, (d) x3 + 6x2 + 30x +25 =0, (e) x3 6x +4 =0, (f) x3+6x +2 =0, (g) x3 +18x+ 15 =0, (h) x33x23x+11 =0, (i) x3 + 3x2 6x + 4 =0, (j) x3+9x 26 =0, (k) x3 +24x-56 =0, (1) x3 + 45x 98 =0, (m) x3 +3x2 -3x-1 =0, (n) x3 6x2 + 57x 196 =0, (o) x3 +3x-2i= 0, (p) x36ix+ 4(1 i)=0, (q) x3 3abx + a3+ b3 =0, (r) x3-3abfgx+f 2ga3+ fg2b3 =0, (s) x3 4x 1 = 0; (t) x3 4x + 2 =0.
*76. Using Cardan's formula, prove that
(x1 x2)2(x1 x3)2(x2 x3)2 = 4p3 27q 2 if x1, x2, x3are roots of the equation x3 + px+ q= O. (The expression 4p3 -27q2 is called the discriminant of the equation x3+px + q = 0.)
20
PART 1. PROBLEMS
*77. Solve the equation (x3 3qx +p3 3pq)2 4(px +q)3=0. *78. Derive a formula for solving the equation x5 -5ax3 +5a2x-2b=0.
79. Solve the following equations:
(a) x4 -2x3 +2x2 +4x-8 =0, (b) x4 +2x3 -2x2 +6x 15 =0, (c) x4 x3 x2 + 2x 2 = 0, (d) x 4 4X3 3X2 +2x-1=0, (e) x4 -3x3 +x2 +4x 6 =0, (f) x4 -6x3 + 6x2 +27x 56 =0, (g) x4 2x3+ 4x2 2x + 3 = 0, (h) x4 X3 3x2 +5x 10 =0, (i) x4 + 2x3+ 8x2 + 2x + =0, (j) x4 +6x3 +6x2 -8 =0, (k) x4 -6x3 +10x2 -2x 3 =0, (1) x4 -2x3 +4x2 +2x 5 =0, (m) x4 x3 -3x2 +x+1=0, (n) x4 x3 4x2+ 4x + 1 =0,
(o) x4 2X3 X2 +2x-1=0,

(p) x4 4X3 2X2 8x +4 =0, (q) x4 2x3+3x2 -2x-2 =0, (r) x4 X3 +2x-1=0, (s) 4x4 4X3 3X2 2x+1 =0, (t) 4x4 4x3 -6x2 +2x+1=0.
80. Ferrari's method for solving the quartic equation x 4+ +ax3+bx2+cx+d=0 consists in representing the left member in the form
(X2 + X +2 )2 [(a 42 + 2
b) X2 +(a 2 A c ) x + ? -a)]. 4
21
Then X is chosen so that the expression in the square brackets is the square of a first-degree binomial. For this purpose it is necessary and sufficient that (c` (2 -; c)2 - 4(4 )(
=o,
that is, A must be a root of some auxiliary cubic equation. Having found A, factor the left member. Express the roots of the auxiliary equation in terms of the roots of the fourth-degree equation. Sec. 4. Roots of Unity 81. Write the following roots of unity of degree (a) 2, (b) 3, (c) 4, (d) 6, (e) 8, (f) 12, (g) 24. 82. Write the primitive roots of degree (a) 2, (b) 3, (c) 4, (d) 6, (e) 8, (f) 12, (g) 24. 83. To what exponent do the following belong: (a) zk = cos 2kn 180 i sin 180 for k = 27, 99, 137 ;
2krc 2k7r
(b) zk = cos 144 2k7t + i sin 144 for k = 10, 35, 60? 84. Write out all the 28th roots of unity belonging to the exponent 7. 85. For each of the roots of unity: (a) 16th, (b) 20th, (c) 24th, indicate the exponent it belongs to. 86. Write out the "cyclotomic polynomials" X (x) for n equal to: (a) 1, (b) 2, (c) 3, (d) 4, (e) 5, (f) 6, (g) 7, (h) 8, (i) 9, (j) 10, (k) 11, (1) 12, (m) 15, (n) 105.
1 +e+ e2 +...+en-1.
*87. Let e be a primitive 2n-th root of unity. Compute the sum
*88. Find the sum of all the nth roots of unity. *89. Find the sum of the k th powers of all nth roots of unity. 90. In the expression (x+ a)m substitute in succession, for a, the m mth roots of unity, then add the results. *91. Compute 1 +2e + 3 e2 + +n en -1, where e is an nth root of unity.
22
PART 1. PROBLEMS
*92. Compute 1 + 4 e + 9 e2 + + n2en -1, where e is an nth root of unity. 93. Find the sums: Ln i n + . . . + (n 1) cos 2 27c n- +2 cos (a) cos (b) sin 277+ 2 sin 4n + . . . + (n 1) sin
(n1)
2(n-1) Tc
*94. Determine the sum of the following primitive roots of unity: (a) 15th, (b) 24th, (c) 30th. 95. Find the fifth roots of unity by solving the equation x5 1=0 algebraically. 96. Using the result of Problem 95, write sin 18 and cos 18. *97. Write the simplest kind of algebraic equation whose root is the length of the side of a regular 14-sided polygon inscribed in a circle of radius unity. *98. Decompose xn 1 into linear and quadratic factors with real coefficients. *99. Use the result of Problem 98 to prove the formulas: n (a) sin 2m 2rc sin 2m . sin (m-1)7
2m 2"1-11; 2m
2"I
2Tc Tr /Mt (b) sin 2m+ 1 sin 2m+ 1 . . . sin 2m +1

n-1
*100. Prove that fl(a + bek )= an + (

k =0
1)n-1b'l
where
Ek= COs 2kir
n
2kTc +i sin n
*101. Prove that

nI
H (d-2ek cos 0+1)=2 (1cos ne),

k= 0
if
Ek = COS
2krc
n
+ sin n
2kTr
102. Prove that

N (t+ con1
k =0
=
k= I
[tn (ek 1 )n]
2k7r where ek = cos -- + i sin
2k7
CH. I COMPLEX NUMBERS
23
*103. Find all the complex numbers that satisfy the condition =xa-' where 5e is the conjugate of x. 104. Show that the roots of the equation X (z a)a + (z b = =0 , where A, t,, a, b are complex, lie on one circle, which in a particular case can degenerate into a straight line (n is a natural number). *105. Solve the equations :
)'
(a) (x+1)m (x 1)'n = 0, (b) (x + i)m (x O'n = 0, n -2 an =__ 0. (c) xn naxn-' c2a2x, 106. Prove that if A is a complex number with modulus 1, then the equation ( I + ix n =A
)'
\ I ix
has all roots real and distinct. *107. Solve the equation cos so + Cn i cos (cp + oc)x + C,, 2cos (q) + 2(x) X2
++e nz cos (cp + not) xn =O.

Prove the following theorems: 108. The product of an ath root of unity by a bth root of unity is an abth root of unity. 109. If a and b are relatively prime, then xa 1 and xb 1 have a unique root in common. 110. If a and b are relatively prime, then all the abth roots of unity are obtained by multiplying the ath roots of unity by the bth roots of unity. 111. If a and b are relatively prime, then the product of a primitive ath root of unity by a primitive bth root of unity is a primitive abth root of unity, and conversely. 112. Denoting by cp (n) the number of primitive nth roots of unity, prove that p(ab)=p(a)cp(b) if a and b are relatively prime. *113. Prove that if n= j,;( where p p2, p, are distinct primes, then p (n) = n(1 Pi
Pa
(1 --) Pk
114. Show that the number of primitive nth roots of unity is even if n> 2. 115. Write the polynomial X (x) where p is prime. *116. Write the polynomial Xi', (x) where p is prime.
,
24
PART 1. PROBLEMS
*117. Prove that for n odd and greater than unity, X2n(x)= = X(x). 118. Prove that if d is made up of prime divisors which enter into n, then each primitive ndth root of unity is a dth root of a primitive nth root of unity, and conversely. *119. Prove that if n= p7' p`P pm kk where pi, P2 , Pk are distinct primes, then Xn(x)= X, (xn") where
17'
=PiP2 Pk, n" =
*120. Denoting by 11(n) the sum of the primitive nth roots of unity, prove that p.(n) =0 if n is divisible by the square of at least one prime number; p.(n)=1 if n is the product of an even number of distinct prime numbers; 1..1.(n)= 1 if n is the product of an odd number of distinct prime numbers. 121. Prove that Ell (d)=0 if d runs through all divisors of the number n,
n'\
*122. Prove that Xn(x)= IZ (xd Cl/where d runs through all divisors of n. *123. Find X(1). *124. Find Xn (-1). *125. Determine the sum of the products of the primitive nth roots of unity taken two at a time. *126. S= 1 + e+ e4 + e9 + ...+ oa where e is a primitive nth root of unity. Find S I.
CHAPTER 2 EVALUATION OF DETERMINANTS
Sec. 1. Determinants of Second and Third Order Compute the determinants: 127. (a) (d) 2 1 3 4 ' (b) 2 1 1 2 (e) , (c) a+p y
cos OC
sin a cos a cos a sin a y + si cx pi
c+di a cdi sin oc

OC
sin oc
(f) sin p cos p

(h) tan cc 1 tan a
1
(g) sin p cos p '

1+ . (1) I 2 + V 3 (k) 2 V3 1 V2
(j) (1
1 logb a 1 logo b a+b ab co 1 ab a+b col 6.)
a+b a+c x-1

x3
b+d I c+d ' 1 x2+ x+1
(m)
(n)
2rc . where co = cos 3 + sin 2 3' (o)
1
7T
where s = cos 3 + sin 3'
26
PART I. PROBLEMS
1 128. (a)
1 (b)
0 1 1
1 0 1
0 1 , - 1 - 1 -1 0
1 1 0
(c)
(e)
a aa a x , (d) -a - a -a x 1 i 1 +i -i 1 0 , 0 1
1
1 1 1
1 1 2 3 3 6
cos S i sin s 7 cos if +i sin 71 sin

it 3
7
TC
(f)
cos
-- - i 3
TC
1 27 . cos - - / sin
cos T - + i sin
27
27
27
cos - i sin 4
1
.
(g)
s , where (,)= cos 3 +i s i n --
(h)
where co = cos -3 Sec. 2. Permutations
27
i sin
27 3
129. Write out the transpositions enabling one to go from the permutation 1, 2, 4, 3, 5 to the permutation 2, 5, 3, 4, 1. 130. Assuming that 1, 2, 3, 4, 5, 6, 7, 8, 9 is the initial arrangement, determine the number of inversions in the permutations : (a) 1, 3, 4, 7, 8, 2, 6, 9, 5; (b) 2, 1, 7, 9, 8, 6, 3, 5, 4; (c) 9, 8, 7, 6, 5, 4, 3, 2, 1. 131. Assuming 1, 2, 3, 4, 5, 6, 7, 8, 9 to be the initial ordering, choose i and k so that: (a) the permutation 1, 2, 7, 4, i, 5, 6, k, 9 is even; (b) the permutation 1, i, 2, 5, k, 4, 8, 9, 7 is odd. *132. Determine the number of inversions in the permutation n- 1, ..., 2, 1 if the initial permutation is 1, 2. n.
CH. 2. EVALUATION OF DETERMINANTS
27
*133. There are I inversions in the permutation al, How many inversions are there in the permutation an, ,_,, c(2, ? 134. Determine the number of inversions in the permutations: (a) 1, 3, 5, 7, ..., 2n- 1, 2, 4, 6, ..., 2n, (b) 2, 4, 6, 8, ..., 2n, 1, 3, 5, ..., 2n - 1 if the initial permutation is 1, 2, ..., 2n. 135. Determine the number of inversions in the permutations: (a) 3, 6, 9, ..., 3n, 1, 4, 7, ..., 3n -2, 2, 5, ..., 3n - 1, (b) 1, 4, 7, ..., 3n -2, 2, 5, ..., 3n - 1, 3, 6, ..., 3n if the initial permutation is 1, 2, 3, ..., 3n. 136. Prove that if a1, a2, a is a permutation with I the number of inversions, then, when returned to its original ordering, the numbers 1, 2, ..., n form a permutation with the same number of inversions I. 137. Determine the parity of the permutation of the letters th, r, m, i, a, g, o, 1 if for the original ordering we take the words (a) logarithm, (b) algorithm. Compare and explain the results. Sec. 3. Definition of a Determinant 138. Indicate the signs of the following products that enter into a sixth-order determinant: a32a43a14a51a66a25 (a) a23a31a42a56a14a65, (b) 139. Do the following products enter into a 5th-order determinant: (a) a13a24a23a41a55, (b) a21a13a34a55a42? 140. Choose i and k so that the product ava 32a4ka25a53enters into a fifth-order determinant with the plus sign. 141. Write out all the summands that enter into a fourth-order determinant with the plus sign and contain the factor a23. 142. Write out all the summands that enter into a fifth-order determinant and are of the form a14a23a3,a4,a5,,. What will happen if a14a23is taken outside the parentheses?
28
PART 1. PROBLEMS
143. With what sign does the product of the elements of the principal diagonal enter an nth-order determinant? 144. What sign does the product of elements of the secondary diagonal have in an nth-order determinant? *145. Guided solely by the definition of a determinant, prove that the determinant
a1 c(2
(43
(X4
P4
C5
P5
( 13 1
12
P3
a1 b1
c1
a,
b2
c2
0 0
0
0 0
0
0 0
0
is zero. 146. Using only the definition of a determinant, evaluate the coefficients of x4 and x3in the expression
2x x 1 x f (x)= 3 2 1 1
147. Evaluate the determinants:
1 1 x 1
2 1 1
(a)
1 0 0 2 0 0 0 0
0 ... 0 0 ... 0 , 3 ... 0 0 ... n
0 (b)
0 .
0 1
0
0
0 ... 1
0 ... 0
1 0
0
(c)
1 a a ... a 0 2 a ... a 0 0 3 ... a 0 0 0 ... n
Note: In all problems, determinants are taken to be of order n unless otherwise stated or unless it follows from the conditions of the problem.
29
148. F(x)=x(x-1) (x-2)...(x n+ 1). Compute the determinants: F(0) F(1) F(2) F(n) F(3) ... F (n + 1) F(2) (a) F(1 ) F(n) F (n + 1) F (n + 2) . . . F (2n) F (a) (b) F' (a) F' (a) F" (a) F" (a) F" (a)
F(n+2) (a)
F'"' (a) pn+ 1) (a) ... F( 2n) (a)
F(n) (a) F<"-") (a)
Sec. 4. Basic Properties of Determinants *149. Prove that an nth-order determinant, each element aik of which is a complex conjugate of ak is equal to a real number. *150. Prove that a determinant of odd order is zero if all its elements satisfy the condition aik + ak, 0 (skew-symmetric determinant).
all a12
151. The determinant
a1
a2n
a21 a 22
is equal to A.
ant an2 a. To what is the following determinant equal
a21 a22 a31 a 32

and
a,
a3
9
an
a,72 ann a12 a,
152. How is a determinant affected if all columns are written in reversed order? *153. What is the sum of al, . . .
E
a, a 2, . . . a2,
an a2 . . . anocn if the summation is taken over all permutations of
w2, ,
30
PART 1. PROBLEMS
*154. Solve the equations:
(a)
1 1 1
x a1 a2
x2 al2
, 2
t..9
. . . xn-1
... an =0 a'2 1-1

. . .
1 an _, an 2 _, . . . where al, a2, an _, are all distinct; 1 1 1 1 1 Ix 1 1 (b) 1 1 1 2 x... . (n-1) x a1 a3 a2 a3 a1a, + a, x a2 a, + a, x a1 1 1 1
= 0;
(c)
a an a
= 0.
a1 a2 a3 . . . a_,+ an x *155. The numbers 204, 527 and 255 are divisible by 17. Prove that 17 divides 2 0 4 5 2 7 2 5 5 *156. Compute the determinant OC 2 (a + 1)2 (oc+2)2 (a + 3)2
p2 Y2
32 157. Prove that b+c c+a a+b b1+c1 c i + a, al +b, = 2
(P + 1 )2 (P +2)2 (P + 3)2 + 1 )2 ("1' + 2)2 (y+ 3)2 2 (6+ 1)2 (8 + 2)2 (8+3) a al. b1 c1
a2 b 2 +a2 a 2 +b2 b2 +c2 c 2 C2 158. Simplify the determinant am + bp an + bq by expancm + dp cn + dq ding it into summands.
CH. 2. EVALUATION OP DETERMINANTS
3l
159. Find the sum of the cofactors of all elements of the determinants:
(a)
a, 0
0
0
a2
0 ... 0 0 ... 0 0 . .. an
(b)
0 0
0 ... 0 0 . . . a,
al 0 0
a 0 ... 0
160. Expand the following determinant by the elements of the third row and evaluate: 1 0 0 1 1 1
a 1
b 1
c 1
1 1 d 0
161. Expand the determinant 2 1 1 1 1 2 1 1 1 1 2 1
x y z t
by the elements of the last column and evaluate. 162. Expand the determinant
al b 0 c 1 d 1
1 1 0 1
1 1 1 0
by the elements of the first column and evaluate. Sec. 5. Computing Determinants Compute the determinants: *163. 13547 28423 13647 28523 164. 246 1014 342 427 543 721 327 443 621
32
PART I. PROBLEMS
165.
3 1 1 1 1 1 1 1 2 1 1 1 1 0 1 1 1 x 1 1 0 0 1 1 2 2
1 3 1 1
1 1 3 1
1 1 1 3
166.
1 1 1 1
169.
1 1 1 2 3 4 3 6 10 4 10 20 1 2 3 4 2 1 4 3
167.
1234 234 1 3412 4123 4 3 2 1
168.
1 1 1 2 3 4 4 9 16 8 27 64 1 3 1 1 1 1 1 4 1 1 1 1 1 1 1 1 5 1 1 6 1 b c 0 1 1 0 x 0 2 2 1 1
3 4 1 2
170.
171.
5 6 0 0 0 1 5 6 0 0 0 1 5 6 0 0 0 1 5 6 0 0 0 1 5
y x +y
172.
1 1 0 a a 0 b c 0 1 x 1 0 x 1 1 1 1 1 1 2 x2 3 3
173.
x y x+ y
175.
x+ y x
x y
174.
0 0 1 1 x 3 3 5 9 x2
1+x 1 1 1 1 1x 1 1 1 1 1 +z 1 1 1 1 1 z
176.
177.
cos (a b) cos (b c) cos (c a) cos (a + b) cos (b + c) cos (c + a) sin (a + b) sin (b + c) sin (c+ a)
33
178.
0 a b c
a 0 d e
bc d e 0 f f 0
*179.
1
1 1
2 0 2 2
3 ... n 3 ... n 0 ... n 3 . .. 0

an a a an+b
xn-1
Xn _1
1
*180. 1 1 1 1 *181. 1 1
x1
x
a2 a1 a2 al +bi a,+b, ... al
a1
x2
x2
a2
xn Xn
1 x1 x
1 1 x1 x1 X2 X2
. .
xn_1 xn
X Xn
xn _1
*182.
1 2 3 . . n-1 1 3 3 . . . n-1 n-1 1 2 5

. . .
1 1
*183.
n 2 3 . . . 2n 3 n- 1 2n 1 2 3
. . .
1 2 2 ... 2
2 2 2 . .. 2 2 2 3 ... 2 2 2 2 ... n
'2. 1215
34
PART I. PROBLEMS
*184.
1 -1 0 0 0
0 0 b1 1-b1 b2 0 -1 1- b2 b, 0 0 0 0 0 0
. . .
0 0 0
0 0 0
. . .
1-b_ 1 bi -1 1 -b.
*185.
a a+h a+2h ... a+(n-1)h 0 0 -a a 0 0 -a a 0 a a 0 . a 0 0 -(a+h) ... (-1)n-1[a+(n-l)h] 0 a 0 a

. ,
*186.
*187.
0 0 a 1 CI C; i C?, ... qr-2 C7,-1 C7, 1 Cl_i C',__ 2 I C!_i . . . C7,..72 qz=1 0 0 ,1 1 0 C; 1 -2 C, 2 ,_.2 C;,-2 . . . CF 1 Cl 0 C2 0 0 CI a1 a2 a3 1 0 ... 0 1 ... 0 x 0 x 0
. . .
1 1 ao *188. ao a1 a2
0 0 an-2 0 0 0 -1 x 3 0 0 -1 0
0 0 an-1
0 0
an
*189.
0 0 ... x a._1 0 0 0 an n n-1 n-2 0 -1 x 1 0 x

. . . . . . -
2 1 0 0 0 0 x 0 -1 x
0 0
0 0
0 0
35
*190. Compute the difference f(x +1) f(x), where
1 1 f (x)= 1
0 2 3
0 0 3
0 0 0
... 0 ... 0 0
. . .
x2 x3 xn
,n+1
1n 1 n+1
... C2 C3 C 2+1 C3 +1 . . .
`-'n+1
Compute the determinants:

*191.
x a1 a2 al x a2 al a2
an-1 1 a_, 1 a_, 1 1 1 a a a a a a x 0 0 0
*192.
x a a ...a a x a...a a a x ... a a a a ... x
al a2 a3 . . . x al a2 a3 . . . an
193.
x a a a
a x a a a, a2 0 0 1
a ... a ... x...
a ... a 0 ... az ... a3 ...
*194.
al 0 0 0 1
0 0 0
0 ... an a 1 ... 1 1 0 ... 0 a, . . . 0 a _ 3 ... 0 0 0

0
*195.
a, 0 0 0 1
a2 0 0 1
0 .. . a_, a 1 ... 1 I a
36
PART I. PROBLEMS
*196. h hx hx2
1 h hx
0 1 h
0 0 1
.. 0 0 . . . 0
.
. . .
hxn hxn-i hxn-2 hxn-3 .
. .
*197.
0 1 1 1 1
1 0 x x x
1 x 0
. . .
. . .
. . .
1 1 x x x x x 0
x ... 0 x ... x
*198.
0 1 1 1 1 0 al + a2 . . . ai+ an 1 a2+ a, 0 a2+ a

. . .
an + a, an+ a2 . . .
0 n 1 n 1 1
*199.
1 1 1
1
2 1 1
1 n
3 1 1 1 1 1 n
. . .
. . . . . .
n1 1 1 n 1
*200.
2
1 1
n 1- 1 n
1_ 1
n '
. . .
1_ 1
n
1 1
n
1 1 1 n a2 a 1
xn3
1 .. n
(order n +1).
*201.
1 a xi, 1
X21 X22
a3 . . . an a2 . . . an-1 a . . . an-2
Xn4
. . .
xnl
xn2
37
*202.
1 2 3 4
2 1 2 3
3 2 1 2
4 3 2 1
... n . . . n- 1 . . . n-2 . . n-3 1

0 0 0 0 0 0
b
n
*203.
n-1 n-2 n-3 . . . b1

b, 0 0 0 b2 0 ... 0 ...
ao a1 a2 an_1 a
b1 b, .
0
0 ... 0 ...
bn-1
*204. 0 a a2 0 . 0 1 2a+b (a+b)2 0 1 2a + 3b (a + 2b)2 . . . 0 0 0 0

0 0 0 0
0 0 0
0 0 0
. . . 2a + (2n 1) b (a + nb)2 1 2a+(2n+1)b
*205.
*206. 1 + x1 Y1 1 + x2 Y1
1
x y 0...0 0 0 0 0 x y
0 y 0 0 0... x y 0 .0 x
+ xiY2 . 1+ xi y + X2 Y2 ... 1+ x2 y
1+ xyi
1 +xnY2 ... 1+ xy
207.
al b1alb2 a1 b a, b1a2 b2 a2 b. ab1 a n b2 ab
*208.
a1-Fx2 a2 -1-x1 1+a 2 -Fx2
an + xi
a+
1+an + x
38
PART I. PROBLEMS
209.
an CC
an-FP
an +1
an+p+1
cc
GC
an+p-1 an+2p-1 a n+p2-1_
an+p (p-1)
an+p (p-1)+1
210. Prove that the determinant A (ai) A (a2) A (an) J2 (ai) f2 (a2) f2 (an)
fn (a1)fn (a2) fn (an)
is equal to zero if fi(x), f2(x), f(x) are polynomials in x, each of degree not exceeding n-2, and the numbers al, a2, ..., an are arbitrary. Compute the determinants:
*211.
1 1 0 0
2 x 0 0
n 1 3 4 . 0 0 ... 0 0 0 0 0
a2 x2 x2 a3 0
X3
0 0 x
an-1 an 0 0
0
x 1
*212. al + xi x, 0
0
0
Xn
xn-1
0 1 1 1
*213.
ao
*214.
a1 a2 x1 0 y2 x2
0 0
y1
0
0
an-1 an 0 0
0 0
1 a1 0
0
1 1 0 ... 0 a2 0
0 an
*215. n! an n 0 0
(n-1)! al. (n-2)! a2 0 x (n-1) x 0 0
an 0 0 x
CH.
2.
EVALUATION
OF DETERMINANTS
39
216.
1 1 1 1
0 al 1 0 0
0 0 0 0 a2 0 1 a3 0 1
1 0 0 0 a4
Write an nth-order determinant of this structure and compute it.
Compute the determinants: 217.

+ 13
1
218.
af3
cc + (3
0 p
. .. . ..
0 0
0 0
1
0
+p ... 0
0 ... 1
0
cc+13
2 1 0 1 2 1 0 1 2 0 0 0
0 ... 0 0 .. . 0 1 .. . 0 0 .. 2
0
*219.
2 cos 0 1 1 2 cos 0 0 0
0 ... 0 1 ... 0 0 ... 1
0 0
2 cos 0 0 0 0
220.
1 0 .. . cos 0 ... 1 2 cos 0 1 0 1 2 cos 0 . . . 0 0 0

*222.
... 2 cos 0
*221.
x
1 0 0
1 x 1 0
0 ... 0 1 0 x ... 0
. . .
X1 Y2 X1 Y2 X2 Y2 X2 Y3
XI Y3 X2 Y3
X1 yn X2 Yn
xi y3
x3 Ya
X3Yn
x3 Yn
0 ... x 1 + al 1 1 1 1 1 + a, 1 1
Yn
x2Yn
xnY n
*223.
1 ... 1 ... 1 + a, . . . 1
1 1 1
. . l+ an
40
PART I. PROBLEMS
224.
1 1
1 1 . x x x
1 . . a2+1 1 1
(21+1 1 1 1
1 an _,+ 1 1 an +1 *225. a1 x x x a2 x x x a3
. . . . . . . . .
x
*226. x1 a2 a3 a, x2 a3 a1 a2 x3
a1
x . . . an
*227.
an-1 an an-1 an a_, a x_, a

xis
a2 b, a3121 xl ai1)2 x2 a3 1 .)2 al b3 a2 b3 x3 al b a2 bn as bn

x3 x3 ... X Xn X
. . .
an b1 an b2 an b3 xn
a2 a3
a1 a2 a3 an-1 *228. X1 in x2 xl xl
X2 m X2
X3 m
x3 . xn x2 x1 229. Solve the equation a an x a2 . . . a1 an-1 al a2 . . . an _1 an _i x an al al X a2 Compute the determinants : *230. a 0 0 . . . 0 0 0 a 0 .. . 0 b 0 0 a .. . b 0

. . .
= 0.
an-1
b 0 0 (of order 2n).
0 0 b a 0 0 0 b 0 ... 0 a 0 b 0 0 .. . 0 0 a
CH
2.
EVALUATION
OF
DETERMINANTS
41
*231. I 1 1 1 1 1 *232.
b na (n 1)a (n 2)a
b 2b a a a
3b . . . (n 1) b 3b . . . (n 1)b a . . . (n 1)b
a ...
2 an an 2
2a
a2 (x a1)2 2 2 (x a2)2 at
. .
2 a?
2 a2
. . . (x ar,)2 a1a a2a
*233.
(x a1)2
a1 a2 a1a
a1 a2 .. . (x a2)2 . . . a2 an
. . . (x an)2
*234.
1 b,
1 0
0
0 b2 1 b, b3 1 1 b, 0 0
b4
0 . . . 1 b
an-1 an a_, an an-1 an
*235.
0 b1
a2
0 b1 b2
b, b2 b1 b2
a3 a4 a3 a4 0 a4 b3 b3 b, b,
0 a b_ 1 0 *237.
*236. 1 1 1 1 2 1 x x 3 2 1 x 4 3 2 1 5 4 3 2 ... n .. . n 1 . . . n 2 . . . n- 3 1
2 x 1
x
3 4 2 3 ... n 1 1 2 . . . n 2 x 1 . .. n 3

x x x x
x x
x.
42
PART I. PROBLEMS
*238.
II
al bl
ai
-1
a2 x n - 2
an _1x an
ao x a o X2
0
X
0 0
bn-i
0 0 0
ao x"-1 ao x"
al xn2
a2 xn-3
a, xn-1 a2 x"-2
an-lx bn
*239. Prove that the determinant a00 xn aoi xn a,o x all a20 X2 a21 x
a02 Xn-2 . . . won
a00 ay, a, . . . ae
= Xn
0
a22
0
. 0
ann
aio aii 0 . . . 0 22 . .. 0 a20 a 21 a ano and a an,,
2 xn-2 . . . ano xn and xn-1 *2.
Compute the determinants: *240. 1 1 1 1 C2 CI

C' n '-
*241.
1 C, 1n 1
CL+1
1 C3
1
C;,
1
Cm i
C?,+1
C 41 . . . CW2
I-2 C m 2-En-El Cm 2 +I CMCm-Fn-1 Cm-Fn
m-F2n-1 Cn
*243. 0 C2 CI 1 *244.
Cn 2
0 0 CI
C3 n
0 0 0
. . CP , -1 C'ci.on +1
Ck
Cfc,71
Cm+I1
Ck m+n
ck m+
. . . C m+I
ck m+ + n ,
cik n+n c m k+ + 1n
CV+,,
C'k n+m-Fi
Crc-1-2m
C7:+2m
k-1-2m-1-1
Ci k n-Fm-1-2 Ci k n +2m-I-1
CT-F3m
*245. 1 1 1 1
0 0 ... 0 C1 0 ... 0 CI C2 ... 0
1
x2
1 xn C;-
CiI. 2. EVALUATION OF DETERMINANTS
43
*246.
1 1 1 1 1
0 1
2
3
n
0 0 2! 3.2
0 0 0 3!
1 x x2
x3 2) ...
xn
n (n 1) n (n 1)(n
*247.
a a a
a+3 2a+3 3a+3
a+23 3a+33 6a -F 43
a+33 4a+63 10a+ wa
a+(n 1)3 ... C,;a+CO ... aFi a+ a ma

.
aCn
+ 8 c7,Tloc+ c,n,18 csn'TIcc+ c42a.. q,;-12 cc+ C'&_2a

*249.
*248.
x y y...y y z x y...y y z z x...y y z z z x y z z z ...z x

250.
a a a . .. a 0 a a a ... 0 a 0 b 0 b b
251.
b b
b b
a, x x . . . x y a2 x . . . x
Y Y Y an
c, a a b c2 a b b c3 b 1 b 1
*253. 1 2
a a a c 1
b 1
*252.
X a a a ... a b a pp... b p a p ... b p (3 a ... b
3 2 3 4 ... 1 3 4 5 ... 2
n 1 2 n1
p p p ... a
44
PART I. PROBLEMS
*254.
a a+h a+2h a+(n-1)h
a+h a+2h a+3h a x2

x
a+2h a+(n-1)h a+3h ... a a+h a+4h a+h

xn 1 Xn2
a+(n-2)h
*256.
255.
1 xn-1
x
x
1
x2
x3
. . .
1
*258.
a b c d
b a d c
c d a b
d c b a
257.
abcdefgh badcfehg cdabghe f d cbahgfe fghabcd fehgbadc g he fcdab hgfedcba

*259.
COS"-1cp, COS"-2cp, cos" 1cp2 COS"-2cp2 COS" -1cpn
a1 a, ... a n a1 x a, ... an
x
a1 a 2 a 3
cos pi 1 cos y2 1 1
COS"-2cpn . . . cos c,o,,
260.
1 sin cp, sin2op,
1 sin cp2 sin2cp2
1 sin cp sin2cp cp
sin"-1p1sin"-1 p2 . . .
261.
(a l)" ... (a n)n an an-1 (a 1)n-1 . . . (a n)n-1 a 1 a-1 1 a n 1
45
262.
(a, + x)" (a, + x)n
(al + x)n-1 al + x (a, + x)n-1 . . a, + x
1 1
263.
(a.+1+ x)" (a+1 + x)n-1. . . an+ + x 1 (2n 1)n (2n 2)n . . . nn (2n)n (2n 1).-1 (2n 2)n ... -1(212)n-1 2n 2 2n 1 1 1 1 arl w1 a, a 1w2 a2 a2 . . . a72 w 4 . . . an 1 x2+ 1 x3+ x2 1 x3+ 1 4 + x3 X3-1 + 1
X, +
n 1
2n 1
*264.
*265. 1 xi+ 1 4 + xi 266.
2 xn + xn
xr iz- l + 4-2 41-1 + 4-2
..4 1-1 1 1 + sin 9 sin 9 + sin2 9
1 1 + sin 9, sin 91+ sin2 9,
1 1 + sin 92 sin 92 + sin2 9 2
sinn -2 pi sinn-1% Sinn-292 + Sinn-192 . . . Sinn-29, 4- Sinn-1p
267.
1
cPl ( x1)
92 (xl)
1
pi (x2) 92 (X2)
1
91(xn) <P2 (x n)
(Pn-i (xi) Pn-i (x2) (Pn-i (xn) xk -1 +a kk. k where clik (x)= x
268.
1 F1(cos 90 F2 (cos cp,)

F, (cos 91)
1 F1(cos 92) F2 (cos P2) . . .
1 F1 (cos pn) F2 (cos Cpn)
Fn_, (cos 92) . . . Fn_,(cos
46
PART I. PROBLEMS
where Fk (X) = aok xk + xk-1+ + akk *269.
1 tx,\
\1/ (x ) 2
...
tx2\
1/ (x2 2) ( ln) x.) ( 2
( x \ 1 x \ ( n x 1) \n, 1 ) " VI - 1 i
1
)= x where ( x k 1.2 ... k *270. Prove that the value of the determinant 1 a1a? . . . a?-1 1 a2 a2 ... 1 an an2
an-1 n
(x-1) . . . (x-k+1)
is divisible by in-1 2n-2 (n 1) for integral al, a2, Compute the determinants: *271. 1 1 1 2 23
22n-1
a.
*272. 3 33 . . . . n na n2n-1
X1 X2
Xn
X1-1
x2-1
X2
.2 X2
xn -1
X
2 X
32n-1 . . .
2 Xi
x; *273. ai
an
n-1
al-1 b 1 ar I b2
ann+ibn+1
al-2b
an-2 b2
4+1 b2 n+ ...
b? a2 bri an+, b4 -1
bn +1
an n +1
274.
sinn-1 alsinn-2 , cos al . . . sin alcos"-2 al cos"-3- al 2sinn-2 2 cos 2 . . . sin 2COS"-2 2 cos"-1OC2
Sinn-1OC,, Sinn-2OCnCOS OC . . .
sin (x COSn -2OC COS"-1OC,,
CH. 2.
EVALUATION OF DETERMINANTS
47
*275.
e+1 e+I
afn-' + a, + a,
+
42-2+ (13
. . . 4+1 + . . . -I- -F a2 -1
a? a2
,,n
2+ n1 1 + a, an_F 12 + an '+1 . . . annIl+an-1 an 21 1 + 1 an n+1
Fl
*276. 1 cos cpo 1 cos cp,
cos cos 2cp1
. cos (n 1) 90 . . . cos (n 1) 91
cp_,
1 cos cp._icos 2(p_1. . . cos (n *277. sin (n + ao sin nao . . . sin ao sin (n + 1) a, sin na, . . . sin a,
sin (n + an sinnan . . . sin an

*278. 1 x2 (x2 1) (x2 1) x?-' (xi 1) xr' (x, 1) *279. 1 1 .. . 1 *280. 1
2 Xi
1 xn (x 1) x,2 , (x 1) (x 1) 1
X2 x Xn
2 Xn
xi x2 . . . x',', 281.
1 1
X2 22 Xi X2
X7-2 Xr 2
n2 . xn
Xi
x2
*282.
1
xn
2 Xn
1 + x, 1 + 1 +x, 1
... 1+ xi ... 1+4
1 + xn 1 + xn ... 1 + xn
xf 1
n
x21 . . . X n-1 4+1 . . xs n +1

x2 1 n . . . xn
48
PART I. PROBLEMS
283.
1 X3 1 x x2 x2 x x3 1 1
284.
x x2 x3 x4
4x3
2x 3x2
3x2 2x x 2x 22 x
2n-1 x
4x3 1
1 1 1 1
2x 4x
y
3x2 9x2
y2
4x3 5x4 16x3 25x4

y3 y4
2y
3y2
Xn
4y3
5y4
*285. 1 1 1 1 1 *286. 1 1 1 1 1 1 1
x2
3x2
32 x2 3n-1 x2
...
(n + 1) xn
(n+ 02 xn (n+
yn
y2
x
2x 22 x 2k--1 x yi Y2 yn _ k
x2 3x2 32 x2
3k-1 x2
xn
nxn n2 xn-1
nk-1 xn-1
Yi
yr`
yn 2 k
Yn n=lic
xn 1 xn 2
C 2n _1
3
x2 *287. 1 x 0 1 C2x 0 0 1 0
1
0
0
y
1
0
y2
...
xn-k
yn 1
yn-2 0 ...
0 0
288. (a) Write the expansion of a fourth-order determinant in terms of the minors of the first two rows.
CH. 2. EVALUATION OF DETETZ MINANTS
49
(b) Compute the determinant 1 0 2 0 2 1 0 2 2 0 1 0
2 1 1
usin the expansion by minors of the second order. (c) Compute the determinant 2 1 0 0 1 2 1 0 0 1 2 1 0 0 1 2
via an expansion by second-order minors. (d) Compute the determinant of Problem 145. Compute the determinants: 1 1 (e) 11 2 3 4 3 6 10 4 9 14 5 15 24 9 24 38 (g) 1 xj. a, a2 a3
2 Xi
0 0 0 0 0 0 0 0 0 1 1 1 9 1 5 1 25 81 1 x3 c1
c2
C3
(f)
a1
0 b2
0
0 c1 0 d2
b1
0 a2 0
0 d1
0
c2
1 0 0 0 x2 0 0 0 1 1 b1 1 x2 x3 b2 x, b3 x? x2 x3
2 X2 0 0 0
(h) X
x1
0
cc pi
0
(3
. . .
(3 Yl
Y2
x2
x
... cc
3 p ...
0
Y.
0 ... 0
(i) Compute the determinant of Problem 230 using the Laplace theorem. (j) Compute the determinant of Problem 171 using the Laplace theorem.
50
PART I. PROBLEMS
(k) Compute the determinant 1 1 1 1 2 3 0 1 1 0 x1 x2

0 xi 2
0 0 1
x8 2 .X3
0 0 1
x4 2 .X4
(1) Let A, B, C and D from the array ( al a2 a,
be third-order determinants formed

b1
C1 C2 C3 dl d2 d3
)
b2 b,
by deleting the first, second, third and fourth columns, respectively. Prove that a1 a2 a3 0 0 0
b1 c1 d1 0
b2 c2 d2 0 0 b3 c3 d3 0 0 0 al b1 Ci dl 0 a2 b2 c2 d2 0 a3 b3 C3 d3
ADBC.
(*m) Compute the fifteenth-order determinant A Al Al Al A Al Al A lA formed (as indicated) from the following blocks:
a x 2a a a 2a x x x\
A=
x x
x \x
0 0
0 0
0 0 2a a
0 0 0 0 0 0 2 1 0 0 0 , A1 = 0 1 2 0 0 a 0 0 0 2 1 2a/ \ 0 0 0 1 2,
1
CH 2. EVALUATION OF DETERMINANTS
51
Sec. 6. Multiplication of Determinants 289. Using the rule for multiplying matrices, represent the following products of determinants in the form of a determinant: (a) 4 1 3 3 1 3 2 5 3 6 1 2 1 2 1 1 1 1 2 1 2 2 2 1 3 3 1 3 1 1 3 4 5 1 2 1 1 2
(b)
3 1 1 2 1 1 1
2
1 1 1 2
(c)
290. Compute the determinant A by multiplying it by the determinant 8: 1 2 3 4 1 0 3 8 (a) A = 1 1 0 13 ' 15 5 2 3 1 2 3 11 0 1 0 2 0 0 1 1 1 0 0 0 1 5 12 9 9 5 6 0 2 3 1 2 3 2 1 1 1 1 1 1 1 2 3 3 1 1 1 1 0 0 1 0 2 1 4 2 1 1 1 1 0 0 0 1
a=
(b) A =
, a=
1 1 1 1
abed bade (c) A = cdab dcba
, s=
52
PART I. PROBLEMS
291. Compute the square of the determinant: 1 1 1 1 1 1 1 1 1 1 1 2 2 1 (a) , (b) 1 1 1 1 2 0 3 1 1 1 1 3 7 1 a (c b

)
1 1 1 9
b
a
c
d a b
d
b a ao, n-i =D.
c d
d
c ao2
292. The determinant aoo ail,

an-1, 0
ao1 an
an -1, 1
an -1, 2
an-1, n-1
What is 'Po (xi) 'Pi (x1) (Po (x2) qh (x2) Soo (x.) 91 (xn)
Pn-i (xi) 9.-1 (x2) (P.-1(xn) where <p i (x)--aoi + x+ + a_1, xn-l? Use the result obtained to find the solution of Problems 265, 267, 268. Compute the determinants : *293. (bo +ao)n (b0 +a1)n (b, + an)n 1 a., 1z 3t:
1
(a)
(b1+ (On (b+ (On (b1+ al)n . . . (bn + a1)n

. . .
(b, + a)n
1 a7
(bn+ a)n
(31
1 a1 P2
1al pn
(b)
1 ce2i pi
0022 pg

1 1 1

rzg p;= ,
C2 13K
1 a2 Pi
1
an
1 as 132 1
Pr il
Oqi N 2
p;,z
1 mn pi
1 an P2
1C4n Pn
53
sin 2 a *294. sin (oc2 sin (a+ *295. So

S2 Sn _1 Sn S,,1
sin (a,+ a2) . . . sin (a,+ a) . . . sin (a, + ) sin 2a2 sin (an+ cc:2)
S2 . . . S,,_i
sin 2a 1
X
s
Sn+1 S+2 S2,, 2
xn
where sk =xlf + 4+ . . . + n m 1 *296. a p m 1 p n b a d c n p 1 m d a b c n m 1 p b a d c c d 1 m n p a b n b a d c m1 b m c d a 1 n p c b a 1 d p n m sin y cos cp sin ? *297. cos cp cos 2y sin 2? 2 cos 2cp 2 sin 2<p cos 3y sin 3(p 3 cos 3cp 3 sin 3y cos 4y sin 4y 4 cos 4cp 4 sin 4cp *298. sin ny n sin ny n cos ny cos ny + 1) (I D sin(n+ 1)cp(n+ 1)sin(n+ 1)y (n cos(n+ 1)cp (n + 1) cos cos (n + 2) y (n + 2) cos (n + 2) y sin (n + 2) y + 2) sin (n + 2) y cos (n + 3) y +3) cos (n + 3) y sin (n + 3) y (n + 3) sin (n + 3) cp *300. *299. a1 a2 . . . an _1 ao 1 . 1 1 1 n-1 e 2 an _1 a0 an-2 e 1 E
1 1
Sn
52n-1
xn
o
en-1
e2 (n-1)
e2n-2 en-1)2
a1
a2
a3 . . . a0
cyclic determinant).
where s = cos
i sin
27`
54
PART I. PROBLEMS
301. Apply the result of Problem 300 to the determinant

x it
z y u z y x 302. Apply the result of Problem 300 to Problems 192, 205, and 255. Compute the determinants:
303.
1
1
0 2 _1 c, , ,=;
1 0
1
cn:
1 1
c, ,,=?
304. 1
C,i_ 1 C
2a 3a2
na"-1 2a
305.
nan-1 2a (n 1) an-2
3a2 4a 3
s sa2 s an s an s a, s an_ 1 s a, sa3 sal
where s=
306.
+ a2 + +an. Or 2
tn-1
Q-1
to-1 Cn-1 n 1
On
C,itfl 3 Cn n-2 t tn 2 3t 2
n
Cr,: 2t
q-3/2
Cr 2t
tn-1
Cisaz -4 t3
OP 2
307.
C,3 i tn-4 C7,-1
p 1 1 1
t o-1 np
1 1 1 1 1 1
ds s s ss / s ss N
1 1 1 1 1-1 1 1
1 1 1 1
1 1 1 1
1 1
CIT. 2. EVALUATION OF DETERMINANTS
55
*308.
rc cos -
2rc COS --
COS
it
cos (11 1)7r

n
cos (n-2)7
/7
COS ---11
(n 1) 7C
2rr
CoS --
37r
Cos --
7r
309.
cos nO cos 20 cos 0 cos nO cos 0 cos (n 1) 0 cos 20 cos 30 cos 0
310. sin a sin [a + (n 1) h] sin (a + h) *311. 12 n2
sin [a + (n 1) h] sin (a + h) sin (a + 2h) sin a sin (a + h) sin [a + (n 2) hi

sin (a + 2h) sin (a + 3h) sin a
22 32 12 22 (11
n2
1)2
12
22 32
312. Prove that
56
PART 1. PROBLEMS
313. Compute the determinant

a1 an a_,
a2
a1 a,, a3
a3
a2
a1 a4 .
an--1 an 2
-
a1
(skew-symmetric determinant). *314. Prove that a cyclic determinant of order 2n may be represented as a product of a cyclic determinant of order n and a skew-cyclic determinant of order n. 315. Compute the determinant
al fan E'v'an-1 [142 a2 a1 a3 a
a2
a1 an- 2
p.a
1..ta 3
1La, ... a,
Sec. 7. Miscellaneous Problems 316. Prove that if

a12 (x) a22 (x) an1(x) an2 (x) a1 (x) a2 (x) . a (x)
then
a;, (x) a'12 (x)
a22
a', (x)
(x) A' (x) = a21

an, (x)
(x) . . . azn(x)
a,,(x) a1 (x)
azn
an2 (x) a12 (x)
an (x)
+ +
an (x) a22 (x)
(x)
and (x) 42(x) . d (x)
CH.
2. EVALUATION OF DETERMINANTS
57
317. Prove that
all + x a12 + x . . . ain x + x a22+ x . . . a2 + x

and
all
a21
ant
a22
a2n
+x
a2 + x . . . ann + x
ant
ann
+x
k=1 i =1
where Aik is the cofactor of the element a,k 318. Using the result of Problem 317, compute the determinants of Problems 200, 223, 224, 225, 226, 227, 228, 232, 233, 248, 249, 250. 319. Prove that the sum of the cofactors of all the elements of the determinant all a12 al.
a21
a22 a2n
and ant ann is equal to 1

a21
1 an
a22
1 a2n aln
ann an-1,n
a12
and
an-1,1
ant an-1,2
Prove the following theorems: 320. The sum of the cofactors of all elements of a determinant remains unaltered if the same number is added to all elements. 321. If all the elements of one row (column) of a determinant are equal to unity, the sum:of the cofactors of all elements of the determinant is equal to the determinant itself. 322. Compute the sum of the cofactors of all the elements of the determinant of Problem 250. *323. Compute the determinant 1)'1 (a i+ b2)-1 (al + bn)-i (a1+b 2 + b2)-1 (a2 + b)-1 (a2 + b1)-1 (a
(a,, bi)-1(a+b3)-1 (an + bn)-1
58
PART I. PROBLEMS
324. Denote by P and Q the determinants

a, 1 a1
0
1
0
0 a_,
0
0 1 au--1 0 0
1 0
0
0
0 1 a2
0 1 0

and
a1 0 0
1 0 0
Pn
0 0
0 0
an-2 1
an-1 1
respectively, and prove that = ao

al+ 1
a s+
as +
1 an
Compute the determinants *325.

lc a0 00 bc a 0 0
326.
pq0 00 2pq 00 0 1 p 0 0 000 pq 0001 p
0 bc 00
000 c a 000 b c
*327. Represent the determinant an + x a12 au a21 a 22 x a2n

ant
ant
a,,,, + x
in the form of a polynomial in powers of x.
59
*328. Compute a determinant of order (2n-1) in which the first n-1 elements of the principal diagonal are equal to unity and the other elements of the principal diagonal are equal to n. In each of the first n 1 rows, the n elements to the right of the principal diagonal are equal to unity and in each of the last n rows, the elements to the left of the principal diagonal are n-1, n 2, ..., 1. The other elements of the determinant are zero.
For example,
1 1 1 0 1 1 123 012 001
1 1 0 3 2
0 1 0 0 3
1 1 0 2 0 1 0 0 0 0 0
2
1 0 0 1
1 1 1 0 0 1 1 1 1 0 1 1 1 1 1
3 4 00 0
2 3 40 0 1 2 34 0 0 1 23 4 0 0 0 0 0 0
Compute the determinants: 1 x *329. n 0

0 330.
0
0
x 2 (n 1)
0 x
x 4 3 0
0 1 x 2n
n-1 0 0
331.
1 00 0 0 2 0 0 0 x n2 x 3 0 0
0
0
00 lx 0 0 0
n(a-1)
0
0 332.
0 2a (n 1)(a 1) x 2 3a
0 in-1
2n 1 2n-1 3n-1
a x-1
0
0
0
0
0 a 1 x n nn -1 (n -F 1)n -1
l)n-i
nu -1 (n + 1 ) n-1
(2n
60
PART I. PROBLEMS
333.
1 1
1 1
1 1
1 1 n+1
1 1 1 n n+1 n+2
2n-1
334. Find the coefficient of the lowest power of x in the determinant ( 1 b (1 -I- X) (1 + JC)a n
(1 + Xyaabi (1 + ab 2 (1 + xrsbn nb 2 (1 + JC)anbn
xy
(1 +
CHAPTER 3 SYSTEMS OF LINEAR EQUATIONS
Sec. 1. Cramer's Theorem Solve the following systems of equations: 336. x1 + x2 + 2x3= 1, 335. 2x1 x2 x3 =4, 3x1+ 4x2 2x3 = 11, 2x1 x2 +2x3 = 4, 4x1+ x2 +4x2 = 2. 3x12x2 + 4x3 = 11.
337. 3x1+2x2 + x3= 5, 2x1+3x2 + x3= 1, 2x1+ x2 +3x3=11.

339. x1 + x2 +2x3+3x4 = 1,
3x1-x2-X3 - 2X4 = -4, 2x1+ 3x2 x3-x4= -6, xi+2x2 +3x3 x4= 4.
338. x1+2x2+4x3=31, 5x1+ x2 + 2x3 =29, 3x1 x2 + x3 =10.
340. xi+2x2 +3x3 -2x4= 6, 2x1 x2 -2x3 3x4 = 8, 3x, +2x2 x3 + 2x4 =4, 2x1-3x2 +2x3 + x4 = 8.
341. xi +2x2 +3x3 + 4x4 =5,
2x1+ x2 +2x3 +3x4 =1, 3x1+2x2 + x3+2x4 =1, 4x1 +3x2 +2x3 + x4= 5.
342. x2 -3x3+4x4 = 5, 2x3 +3x4= 4, 5x4 =12, 3x1+ 2x2 4x1+ 3x2 5x, =5.
x1
62
PART I. PROBLEMS
343. 2x1 - x2 + 3x, + 2x4 = 4, 3x1+ 3x2 + 3x3 + 2x4 6, 3x1- X2 X3 + 2X4 = 6, 3x1 - x2 + 3x3 - x4 =6. 344. x1+ x2 + x3 + x4 =0, x1+ 2x2 + 3x3 + 4x4 =0, xi + 3x2 + 6x8 +10x4 =0, x1+ 4x2 + 10x3 + 20x 4 = O. 345. xi+ 3x2 + 5x3 + 7x4 = 12, 3x1 +5x2 +7x3 + x4 = 0, 5x1 +7x2 + x3 + 3x 4 = 4, 7x1 + x2 + 3x3 + 5x4 = 16.
346. xi +x2. +
X3+ X4+ X5 = 0,
xi, -x2 + 2x3 - 2x4 + 3x5 = 0, x1+ X2 + 4x3+ 4X 4 + 9X5 = 0, x1 -x 2 + 8x3 - 8x4 + 27x5 = 0,
X1+X2 16X3+ 16X4+81X5 =0.
347. xi+ 2x2 + 3x3 + 4x4 =0,

+ X2 + 2X3 + 3X4 = 0,
x1+ 5X2 + X3 + 2X4 = 0,
x1 + 5X2 + 5X3 + 2X4 = 0.

348. xi + x2 + x3 + x4 =0,
= 2, = -2,
X2+ x3+ x4+ X5=0,
x1+ 2X2 -1-3X3

X2 + 2X3 + 3X 4
x, + 2x 4 + 3x, = 2.
349. x1+4x2 +6x3 +4x4 + x5 =0,

x1+ X2 +4X3 +6X4+4X5=0,
4x1+ X2+ X3 + 4X4 -1 -6X5 = 0,

6x1+ 4X2 + X3+ X4 + 4X5 =0,
4x1+ 6X2 + 4X3 + X4+ X5 = 0.
CH. 3. SYSTEMS OF LINEAR
EQUATIONS
63
350. 2x1+ x2 + x3 + x4 + x5=2, x1+2x2 + x3+ x4 + x5=0. x1+ x2 +3x3+ x4 + x5 =3, + x2+ X3 4X4 = 2, x1+ x2 + x3 + x4 +5x5 =5. 351. x1+2x2 +3x3 +4x4 + 5x5 = 13, 2x1+ x2 +2x3 +3x4 +4x5=10, 2x1+2x2 + x3 +2x4 +3x5 = 11, 2x1+2x2 +2x3 + x4 +2x5= 6, 2x1+2x2 +2x3+2x4 + x5 = 3.
352. XI 1-2X2 3X3 4X4 x5 = 1, 2x1 x2 + 3x3-4x4 + 2x5= 8, 3x1+ x2 x3+2x4 x5 =3, 4x1+3x2 +4x3 +2x4 +2x5 = 2,
X1 x2 X3 2X4 3X5 = 3.
353. 2x1 3x2 + 4x3 3x4 =0, 3x1 x2 +11x3-13x4 = 0, 4x1+ 5x2 7x3 13x1-25x2 + x3+ 11x4 =0. Verify that the system has the solution x1=x2 =x3=x4=1 and compute the determinant of the system. 354. Prove that the system ax+by+cz+dt=0, bx ay+ dz ct =0, cxdyaz+bt=0, dx+ cybz at =0 has a unique solution if a, b, c, d are real and not all zero. Solve the following systems of equations:
355. ocx1+ocx2 +
+ocx,,_1+ px = a, ocxi+ ocx2+ ...+13xn _ i +ocx = an --1,
(3x1+ocx2 + ...+ can _1+ ocx= a, where cc p.
64
PART I. PROBLEMS
356.
u,
b2 f31 X1
xi
u, N2
b2 (30(
+ + "xn = 1, by P nn
+
x1 4. X 2 4.
4, 2
Xn a
x2 +
bn f31
bn [32
bn (3n
where b1, b2, 357. x1

X10(1
b, (31, p2,
+x2
X20(2
P are all distinct.
+ ...+x
+
= 1,
t, +xan =
xnan n-1 = tn-1
x10(7 1 + x2 4-1
where 0(1, 0(2, , (Xn are all distinct.

358. x, + x2 0(1+ + xnajI = ui,
xi + x2a2 xnce2z-1= u2,
xi
+ x2an +
=u
where al, 0(2i , 359. x,
are all distinct.
+x2 +x2a2
+ ..+xn
.
=
= U2,
+ . +xnan
x10(1-1 + x24-1 +
xnccn-1 = un
where 0(1, 0(2i , an are all distinct. 360. 1+ xl + x2 + + x=0, 1+2x, +22x2 + ... +2 x=0, 1 +nxi+n2x2+ ...+nn x=0. Sec. 2. Rank of a Matrix
361. How many kth-order determinants can be formed from a matrix with m rows and n columns? 362. Form a matrix with rank equal to (a) 2, (b) 3.
CH. 3. SYSTEMS OF LINEAR EQUATIONS
65
363. Prove that the rank of a matrix remains unaltered if: (a) rows and columns are interchanged; (b) the elements of a row or column are multiplied by a nonzero number; (c) two rows or two columns are interchanged; (d) multiples of the elements of one row (column) are added to elements of another row (column). 364. The sum of two matrices having the same number of rows and columns is a matrix whose elements are the sums of the corresponding elements of the matrices being added. Prove that the rank of the sum of two matrices does not exceed the sum of the ranks of the matrices added. 365. How is the rank of a matrix affected by adjoining (a) one column, (b) two columns? Compute the rank of the following matrices: 367. 366. ( 75 0 116 39 0 4 10 1\ 0\ 171 69 402 123 45 4 8 18 7 301 0 87 417 169 10 18 40 17 114 46 268 82 30 / 1 7 17 3/ 369. 368. / 14 12 6 8 2\ 2\ / 2 1 11 6 104 21 9 17 I 1 0 41 5/ \ 7 6 3 4 1 4 56 11 \ 35 30 15 20 5 / 21 56 370. 1 00 010 00 1 1 23 456 372. 2 1 1 1 3 1 1 1 4 1 1 1 1 23 1 1 1
3. 1215
1 4\ 2 5 3 6 14 32 32 77 / 1 1 1 5 4 1 /
371. 1 2 3 1 1 2 \ 21 1 0 2 2 2 5 8 4 3 1 0 1 2 7 5 6 1 1 2 1 / \ 1 1 373. / 1 1 2 3 4\ 0 2 1 1 2 1 2 1 1 3 1 5 8 5 12 13 / 8 9 \ 3 7
66
PART I.
PROBLEMS
374. 1 3 1\ 72 0 3 1 2 1 3 4 2 \4 -3 1 1/ 376. /001 010 000 1 1 1 1 34 1 23 234 378. / 1 0 1 1 1 0 0 1 1 001 \010 380. /21 2 1 2 3 1 0 1 2 4 1 1 3 4 2 1 2 1 2 2 1 2 6 1 1 0 3 1 8 / 0 0 0 1 1 0\ 0 0 0 1 /
375. / 3 21 2 0 1 \ 4 1 0 3 0 2 2 1 2 1 1 3 3 1 3 9 1 6 \ 3 1 5 7 2 7 / 377. / 1 1 20 0 0 1 12 0 1 0 1 0 2 1 1 00 1 2 0 0 1-1 \ 1 1 0 1 1 379. ,2 (0 2 0 0 1 1 1 2 0 0 0 0 1 2 1 2 0 1 0 1 \ 1 1 2 1 2/
0 0 1 1 5 4 5
0 \ 0 0 1 1 5 6/
Sec. 3. Systems of Linear Forms 381. (a) Write two independent linear forms. (b) Write three independent linear forms. 382. Form a system of four linear forms in five variables so that two of them are independent and the others are linear combinations of them.
67
Find the basic dependences between the forms of the system: 383. y1=2x1 +2x2 +7x3 - X 3/2=3x1 x2+2x3+4x4,
y3= + X2 3X3 4- X4.
384. y1=3x1+2x2 5x3 + 4x4,

y2 = 3X1 - x2+ 3x3-3x4,
y3=3x1+5x2 -13x3+ 11x4. 385. y1=2x1+3x2 -4x3 x,,

2X2 + X3 + 3X4, Y2= X1 -
y3=5x1-3x2 x3+8x4, y,---3x1+8x2-9x3-5x4. 386. y1=2x1+ x2 x,+x,

2X2 + X3 -X4, Y2= y3= X1+ X2 2X3 + X4.
387.
388.
yl = x1+2x2 +3x3 + x4,

Y2 = 2X1 3X2 4-X3 4- 2X4,
y3=3x1+ x2+2x3-2x4, 4X2 2x,+5x4. y4=
y1 =2x1+ x2, y2 = 3x1 +2x2, y3= x, +x2, y4= 2X1+ 3X2.
389. y1=x1+ x2 + x3 + x4 +x5,

y2 = 2X2 3X3 4X4 X3,
y3=x1+3x2 + 6x3 +10x4 +x5, y4=x, + 4x2+ 10x3+ 20x4 +x5.

390. 391.
yi= x1+2x2+3x3-4x4,
)12 = 2X1 - X2 + 2X3 5X4,
y1 =2x1+ x2 -3x3, 5x3, y2=3x1 + x2 y3 - 4Xi 2X2 - x3,
y3 =2x1 x2 +5x3-4x4, y4 = 2xi+ 3x2 4x, + x4.
Y4 =X1
-7x3.
68
PART I. PROBLEMS
392. y1=2x1+3x2 +5x3 4x4 + x5,

Y2= - X2 + 2X3 + 3X4 + 5X5,
y3-= 3x, + 7x2+ 8x3 1 1x4 3x5,

y4= X1- X2+ X3-2X4 + 3X5.
393.
x2 + 3x3+ 4x4 x5,

y2= + 2X2 - 3x3+ X4 4-2X5, 5X2 + 12X3+ 1 1X4 -5X5, y3 = 5X1-
y4 = X1-3X2 + 6x3+ 3X4 - 3X5. 394. yi = x1+2x2 + x3 -2x4 + x5,

y2 = 2X1 - X2+X3 + 3X4 + 2X5 , y3- X1-X2 + 2X3 - X4 + 3X5, y4 = 2X1+ X2 - 3X3 + X4 - 2X5, y5= x1 -X2 + 3X3 - X4 + 7X5.
395. y1=4x1+3x2 x 3+ x4 x5,

y2 =2X1+ X2 - 3X3 + 2X4 - 5X5, X4 - 2X5, y3 = x1-3x2 + +6x5. +2x3 -2x4 = x1+5x2 y4
396. yi = x1 +2x2 x3+ 3x4 x5 +2x6,

Y2 = 2X1-X2 + 3X3 - 4x4+ X5- X6, Y3 = 3X1+ X2 -X3+ 2X4+ X5 + 3X6 ,
y4- 4X1 - 7X2 + 8x 3 - 1 5X4 + 6X5 -5x6,

y5 = 5X1 + 5X2 -6X3+ 11X4 +9X6.
397. y,= x1 +2x2 + x3 3x4 +2x5,
y2 =2x1+ x2 + x3 + x4 -3x5,
y3= + X2 + 2X3 + 2X4 2X5, Y4 = 2X1 + 3X2 - 5X3 - 1 7X4 + X.X5.
Choose X so that the fourth form is a linear combination of the other three.
Sec. 4. Systems of Linear Equations 398. Solve the system of equations
x1-2x2 +x3+ x4 =1, x1-2x2 +x3 x4 = 1, x1-2x2+x, + 5x4= 5.
CH. 3. SYSTEMS OP LINEAR EQUATIONS
69
399. Choose X so that the following system of equations has a solution:

2X1-X2+ X3+ x4= 13 x1+2X2 -x3+ 4X4 = 2,
x1 +7x2 -4x3 + 1 lx4 = Solve the systems of equations: 401. 400. 2x1+ x2+ x3= 2, x1+ x2 -3x3 = 1, x1+3x2 + x3 =5, 2x1+ x2 -2x3 =1, x,+ x2 +5x3 = 7, x1+ x2 + x3 =3, 2x1+ 3x2 3x3=14. x1+2x2 -3x3 =1. 402. 2x1 x2 + 3x3 =3, 3x1+ x2 5x3 =0, 4x1 x2 + x3 =3, x1+3x2 -13x3 = 6. 403. x1+ 3x2+2x3=0, 2x1 x2 +3x2 =0, 3x1 5x2 +4x3 =0, xi+ 17x2 +4x3 = O.
404. 2x1+ x2 x3 + x4 =1, 3x1 2x2 +2x3 3x4 = 2,

5x1+ X2- X3+ 2X4 = -1, 2x1 x2+ X3 - 3X4 = 4.
405. 2x1 x2+ x3 x4= 1 , 3x4 =2, 2x1 x2 x3 + x4 = 3, 3x1 2x1+ 2x2 2x3+ 5x4 = 6. 407. x1+2x2 +3x3 +4x4 =11, x2 -x3+ x4= -3, 2x1+3x2 +4x3+ X4 =12, 3x4 =1, 3x1+4x2 + x3+2x4 =13, x1+3x2 7x2 +3x3 + x4= 3. 4x1+ x2 +2x3 +3x4 =14. 409. 408. 3x1+ 4x2 5x3 + 7x4 =0, 2x1+ 3x, x3 + 5x4= 0, 3x1 x2 +2x3 -7x4 =0, 2x1 3x2 + 3x3 2x4 =0, 4x1+ 1 lx2 13x3+ 16x4 = 0, 4x1+ x2 -3x3 +6x4 =0, x1-2x2 +4x3 -7x4 =0. 7x1- 2x2+ x3+ 3X4 = 0. 406. x1-2x2+3x3-4x4=4,
70
PART 1. PRO13LEMS
5 =0, - 3x4 - x 410. x1+ x2 x1 - x2 + 2x, - x4 =0, 4x 5 =0, 4x1- 2x2 + 6x3+ 3x4 2x1+ 4x2 - 2x3+ 4x4 - 7x5= 0. 411. x1 + x2 + x3 + x4 + x5 =7, 3x1+2x2 + x3 + x4 -3x5 = -2, x2 + 2x3+ 2x4+ 6x5 = 23, 5x1+ 4x2 + 3x3+ 3x4 - x, = 12. 4+ 412. x1- 2x2 + x3 -x
X5 = 0,
2X1+ X2 -X3 4- 2X4 - 3X5 = 0, 2X2 - X3 + X4 - 2X5 = 0, 3X1 5X2 X3 - 2X4 2X5 = 0. 2X1 -
413. xi. - 2x2 +
x3+ X4 -X5 = 0,
+ X2- X3 -X4+ X5 = 0, 7X2 5X3 5X4 5X5 =0, 3x1 X2 - 2X3 + x4 -X5 = 0
414. 2x1+ X2 -X3-X4 + X5 = 1 9 x1- x2 + x3 + x4 - 2x5 = 0, 3x1+ 3x, - 3x, - 3x4+ 4x5= 2, 4x1 + 5x2 - 5x3- 5x4+ 7x5= 3. 415. 2x1 2x 2 + x3 - x4 + x5 =1, + 2x2 - x3 + x4 - 2x5 = 1, 4x1- 10x2 + 5x3 - 5x4 + 7x5= 1, 2x1- 14x2 + 7x3 - 7x4+ 11x5 = -1. 416. 3x1 + X2 - 2X3 x4 -X5= 1, 2x1 - x, + 7x3 - 3x4+ 5x5= 2, x, + 3x2 - 2x3+ 5x4 - 7x5= 3, 3x1- 2x2 + 7x3- 5x4+ 8x5 = 3. - 3x4 +2x5 = 1, 417. x, + 2x2 X4 - 3X5 =2, x1 - X2 - 3X3 + 2x1- 3x2 + 4x3 - 5x4 + 2x5 = 7, 9x1- 9x2 + 6x3- 16x4 + 2x5= 25.
71
=1, 418. xi+ 3x2 + 5x3 4x4 x1 + 3X2 2X3 -2X4 + X5 = -1, x1-2X2 4- X3- X4-X5 = 3, 4x2 + X3+ X4 -X5= 3, x1+2x2 + x3 x4 +x5 =-1. 419. xi+ 2x2 + 3x3x4= 1,
3X1+2X2 + X3 -X4 = 1, 2x1+3x2+ X3+X4= 1, 2X1 + 2X2 +2X3 -X4 =
5x1+5x2 +2x3
=2.
420. x1-2x2 +3x3 -4x4 +2x5 = 2, x1+2x2 x3 x5= 3, x1 x2 +2x3 -3x4 = 10,
X2 -X3+ X4 - 2X5 = -5,
2x1+3x2 x3 + x4 +4x5 = 1. 421. The system of equations
ay+bx=c, cx+az=b, bz+cy=a

has a unique solution. Prove that abc00 and find the solution. Solve the following systems of equations: 422. Xx+ y+ z=1, x+Xy+ z=X, x+ y+Xz=X2. 424. x+ ay+ a2z = x+by+b2z=b3, x+cy+c2z=c3. 426. ax+ y+z=4, x+ by+z=3, x+2by+z=4. 423. Ax+ y+ z+ t=1, x+Xy+ z+ t=X, x+ y+Az+ t=X2, x+ y+ z+Xt=X3 425.
x+ y+ z=1, ax+b y+c z=d, a2x+b2y+c2z=d2.
427. ax+ by+ z=1, x+aby+ z=b,

x+ by+az=1.
72
PART .1. PROBLEMS
428. coc+ y+ z=m,
429. x+ ay+ a2z=1,
x+ocy+ z=n, x+ y+ocz=p.

430. (X+3)x+
x+ ay+ abz=a, bx+a2y+a2bz=a2b.
2z= X, y+ z=2A, Xx+(X-1)y+ 3(X+ Dx+ Xy+(X+3) z=3. Xx+Ay+ (A.+1)z=-X, Xx+Xy+ (X-1)z=X, (X+1) x+Xy+(2X+3)z=1. 3kx+(2k+1) y+(k+ z=k, (2k-1) x+(2k-1) y+(k-2) z=k+1, (4k-1) x+ 3ky+ 2kz=1. by+ 2z=1, ax+(2b-1) y+ 3z=1, ax+ by+(b+3) z=2b 1.
431.
432.
433. ax+
3mx+ (3m 7)y+ (m-5) z = m-1, (2m-1)x+(4m 1)y+ 2m z=m+ 1, 4mx + (5m 7)y + (2m 5)z =0. (b) (2m + 1)x my+ (m+1)z=m-1, (m 2)x + (m I) y+ (m 2)z =m, (2m 1)x + (m 1) y + (2m 1)z = m. (c) (5A+ 1)x+ ay +(4A+ 1)z= +A, (4X 1)x+(X y+(4X 1)z= 1, 2(3X+ 1)x+ 2Xy + (5X + 2)z = 2 X.
435. (a) (2c + 1) x
434. (a)
cy (c+1)z=2c, 3cx(2c 1) y(3c 1)z=c + 1, y 2cz = 2. (c + 2) x (b) 2(A + 1)x + Xz= X+4, 3y+ (4X-1)x+(X+ y+(2X-1) z=2X+2, (5X-4)x + (A + 1) y + (3X 4) z= X-1. (c) dx + (2d 1)y + (d +2)z =1, (d-1)y+ (d-3)z=1+d, dx+(3d-2)y+(3d+1)z=2d.
73
2ay+ (3a+1)z=1, (d) (3a-1) x+ 2ay+ (3a+1)z= a, 2ax+ (a+1) x+(a+1) y+ 2(a+1)z=a2.
436. Find the equation of a straight line passing through the points M1(x1, Y1), M2(x2, Y2) 437. Under what condition do the three points M3(x1, M2(x2, Y2), M3(x3, yo) lie on a straight line? 438. Under what condition do the three straight lines aix+ + =0, a2x+b2y + c2 = 0, a3x+b3y+c3=0 pass through one point? 439. Under what condition do the four points Mo(xo, yo), M2(x2, Y2), M3(x3, Y3) lie on one circle? Mi. (x1, 440. Write the equation of a circle passing through the points M1(2, 1), M2 (1, 2), M3(0, 1). 441. Find the equation of a quadric curve passing through the points MI (0, 0), M2 (1, 0), M3 ( -1, 0), /114(1, 1) and M5 (- 1, 1). 442. Find the equation of a third-degree parabola passing through the points M1(1, 0), M2(0, -1), M3 (- 1, -2) and M, (2, 7). 443. Form the equation of a parabola of degree n y=aoxn+ +aixn-l+ ...+ a passing through the n+1 points M, (x, yo), M1 (x1, yi), M2 (x2, Y2), Mn (xn, Yn) 1, Yi, 444. Under what condition do the four points M1 (x M2 (x2, Y2, z2), M3 (x39 Y39 z3), M4 ()C49 Y49 z4) lie in a single plane? 445. Form the equation of a sphere passing through the points Mi.(1, 0, 0), M2 (1, 1, 0), M3 (1, 1, 1), M4 (0, 1, 1). 446. Under what condition do the n points M1 (x1, yi), M2 (X29 Y2), M3(x3 ,y3), 9 Mn(Xn, yr) lie on a single straight line? 447. Under what condition are the 11 straight lines alx+b1Y+ +c1=0, a2x+b2y+ c2 =0, ..., anx +kJ + c=0 concurrent? 448. Under what condition do the n points M1(x1, Yi, z 1), M2 (x2, Y2, z2), 9 Mn Oct, yn ZO lie in one plane and under what condition do they lie on one straight line? 449. Under what condition do the n planes A,x+.13,y+Ciz+ +D1=0 (i=1, 2, ..., n) pass through one point and under what condition do all these planes pass through a single straight line? 450. Eliminate x1, x2, ..., x_1from the system of n equations:
iXn -1+ ain =0, a11x1 +a12x2+ +al, n anxi +a22x2+ + a2, n- iXn -1+ a2n = 0,
anix1 +a2x2 + ... +an, n -1x,-1 -1- ann=0.
74
PART I. PROBLEMS
451. Let
1) xt = an, 2 = an;
) . (
,12) '9219 ,(2) = ` ,22 , 219 9^
(1)
(m) = amn X(r ) Mm i; X2 (m) = Km2; . . .; X n
be m solutions of some system of homogeneous linear equations. These solutions are termed linearly dependent if there exist constants cl, c2 cmnot all zero, such that clap+ c2a2i + ... + cmcc,n, = 0 (i= 1, 2, .. n). (2)
= cm= 0, If the equations (2) are only possible when c1=c2= then the solutions are termed linearly independent. Let us agree to write the solutions as rows of a matrix. Thus, the system of solutions of (1) is written in matrix form as
7 an c(12
a21
CX22 amt amt
C(ln
C2n
=A.
chm
Prove that if the rank of matrix A is r, then the system (1) has r linearly independent solutions and all other solutions of (1) are linear combinations of them. 452. Prove that if the rank of a system of m homogeneous linear equations in n unknowns is equal to r, then there exist n r linearly independent solutions of the system, and all other solutions of the system are linear combinations of them. Such a system of nr solutions is termed a fundamental system of solutions. 453. Is ( 1 2 1 2 0 0 1 2 0 1 0 1 1 1 3 2 0 0 0 0 a fundamental
system of solutions of the system of equations x1+ x2 + x3+ x4 + x5 =0, 3x1+2x2 + X9+ X4 3X4 = 0, 2x3 +2x4 +6x5=0, 5x1+4x2 +3x5 +3x4 x5=0?
75
454. Write a fundamental system of solutions of the system of equations of Problem 453. 455. Is ( 1 2 1 0 0 0 I 1 0 0 6 4 0 0 2 a fundamental
system of solutions of the system of Problem 453? 456. Prove that if A is a rank r matrix that forms a fundamental system of solutions of a system of homogeneous linear equations, and B is an arbitrary nonsingular matrix of order r, then the matrix BA also forms a fundamental system of solutions of the same system of equations. 457. Prove that if two matrices A and C of rank r form fundamental systems of solutions of some system of homogeneous linear equations, then one of them is the product of some nonsingular matrix B of order r by the other; that is, A=BC. 458. Let / an c(21
\ ar1
C(12 Cqn
be a fundamental system of so-
c(22 M 2n art Crn
lutions of some system of homogeneous linear equations. Prove that

XI C11 11+ C2C(21+ + Cr(Xrl
X2 = CICC12 C2C422 + + Crar2I
x= ciocin + coc2-P
crocn,
is the general solution of this system of equations, i.e., that any solution of the system may be obtained from it for certain values of el, c2, cr, and conversely. 459. Write the general solution to the system of Problem 453. 460. Verify that (11 1 7) is a fundamental system of solutions of the system of Problem 403, and write the general solution. 461. Write the general solutions of the systems of Problems 408, 409, 410, 412, 413. 462. Knowing the general solution of the system of Problem 453 (see the answer to Problem 459) and the fact that x1= 16, x2 =23, x3=x4 =x5 =0 is a particular solution of the system of Problem 411, find the general solution of the system of 411. 463. Write the general solutions of the systems of Problems 406, 414, 415.
CHAPTER 4 MATRICES
Sec. 1. Operations on Square Matrices

464. Multiply the matrices :
73 2 1)' (b) \ 6 1 ). (1 1 1 (a) ( 3 2 1-1 1 311 1 , (c) 2 1 2 2 1 1 0 (1 2 3) 1 1 2 (d) 2 4 (3 6 1 2 (e) 0 1 (3 1 3 6 9) 1
\ 7 1 )' 1j. 2 3 2
--2 2 2
4
4 ,
4 1 2 , 1
1) 1 2 1) 2 3 1 0 - 0 1 2 . 1 1 2 1 3 1 if 1 a c a b c (f) ( c b aI b b. 1 1 1 1 c a 465. Perform the following operations :
2 1 1 2 2 1 )3' (a) (3 1 0 , (b) ( 1 3 (c) ( 3 4 0 1 2 sine \ n /cos so (d) sin p cos p)
2)5 2 '
CO 1 i' \ (e)
1
*466. Find Ern (
n > co
) n
0,
-
, where a is a real number.
CH. 4. MATRICES
77
467. Prove that if AB= BA, then
(a) (A + B)2= A2 + 2AB + B2, (b) A2 B2 =(A + B) (A B),
(c) (A + B)" = A" + An-1B+ . . . + Bn

468. Compute AB BA if: 2 (a) A= (1 1 2 1 2 1 2 , 3 4 B= 4 1 1 2 2 1 0 ; 1 1 2 5
(2 1
A= (b)
1 1
1 2
0 2 , 1
B=
3 3 3
2 4 . 1
469. Find all matrices that commute with the matrix A : (a) A =
2\
1'
(b) A =
71 1\ \ 0 11
1 0 0 (c) A = (0 1 (). 3 1 2
470. Find f (A): 1 1 3 1 2 ; (a) f (x)= x2 x 1 , A= (2 1 1 0 1\ (b) f (x)= x2 5x+3, A =( 2 3 3/'
471. Prove that every second order matrix A = c

-
(a b\ satisd
fies the equation
x2 (a+ d)x + (ad bc)= 0.

472. Prove that for any given matrix A there is a polynomial f(x) such that f(A) = 0 , and that all polynomials with this property are divisible by one of them. *473. Prove that the equation AB BA=E is impossible. 474. Let Ak =O. Prove that (E A)-= E + A + A2+ . . . + Ak
78
PART I. PROBLEMS
475. Find all second-order matrices whose squares are equal to the zero matrix. 476. Find all second-order matrices whose cubes are equal to the zero matrix. 477. Find all second-order matrices whose squares are equal to the unit matrix. 478. Solve and investigate the equation XA =0, where A is the given matrix and X is the second-order matrix sought. 479. Solve and investigate the equation X2=A, where A is the given matrix and X is the desired second-order matrix. 480. Find the inverse of the matrix A: 2 \' (a c b d), (a) A=(1 A= 2 5) (b) 1 2 3 (c) A= (0 1 2 , (d) A= 0 0 1 , 1 3 5 7\ 0 1 23\ 0 0 1 2 0 0 0 1/ 1 1 1 1 1 1 1 1 1\ 1 1 1l
'
2 2 3 (i il (e) A= 1 1 0 , (f) A= 1 1 2 1 \1
72 1 3 2 (g) A= 1 1 \2 1 1 1 1 e
2
0 0 3 2 1 e2
e4
0 0 , 4 3
0 1 (h) A= 1
1 en 1
e2n 2
1
0 1 1
1 ... 1 1 ... 1 0 ... 1

1 ... 0
(1) A= 1
1
_
27t
n
e 2n 2
2n
e(n 1)2
where e = cos
+ i sib 11
CH. 4. MATRICES
79
(J) A =
2 1 0 0
1 2 1 0
0... 1... 2...
0\ 0 0
0 . . . 1 2,
3 5 7 . . . 2n 1 1 3 5 . . . 2n 3 1 2n 1 (k) A= 2n 3 2n 1 1 3 . . . 2n 5 3 5 7 9... 1
(1) A =
'1 0 0 . . . 0 c1 \ 0 1 0 . . . 0 c2 0 0 1 . . . 0 c3 0 0 0 . . . 1 c b a, b, b2 b, 1 x 0... 0 1 x ... 0

\ a,
0\ 0 x an
1
(m) A = 0 0 ...1 a4 1
al
/1 ,
(n) A = 1 1 1 ... 1 +
/ (o) Knowing the matrix B-1, find the inverse of the bordered ( B U\. matrix V a 481. Find the desired matrix X from the equations: 2 5 =( 6 \ (a) ( 3) X 2 1j '
80 2 (b) X (1 1 /1 1 1 0 1 1 (c) 0 0 1
\
PART I. PROBLEMS
1 1 1
1)
1 1 3 0= 4 3 2 , 1 1 2 5
1\ '2 1 0 0\ 1 2 1 0 1 1 -X= 0 1 2 ... 0 ,0 0 0 ... 2/ 2\ 3/ 1 1 0 0 0 0 1 4 1) , 0 1 1 1 1 1 0 1 X. 1 1 0 0 0 0 0 0\ 0 0 .. 1/ 0 0 0
0 ... 1/
(d) (2 3 2 1 ).X.( -3 5 1 1 0 0 1 1 1 1 . 0... 1...
(e)
0 0 ... 1
11
0 0 0 21 0 1 2 ... 0 0
0 \0 /2 (f) 1\ 1)
0 ... 2 1 0 ... 1 2 (0 1 0 1)
2 1\ (2 1) ; (g) x(2
482. Prove that if AB=BA, then A-1B=BA-1. 1 2 1 +x 483. Compute p (A), where p (x) = , A=( 2 1). 484. Find all the second-order real matrices whose cubes are equal to the unit matrix. 485. Find all the second-order real matrices whose fourth powers are equal to the unit matrix. 486. Establish that there is an isomorphism between the field of complex numbers and the set of matrices of the form a b for real a, b. (b a)
CH. 4. MATRICES
81
487. Establish that for real a, b, c, d, the matrices of the form
a+bi c+di
\c +di a bi
constitute a ring without zero divisors.
488. Represent (4+ b? + c? + d?) (4+14+ c3+ c13) as a sum of four squares of bilinear expressions. 489. Prove that the following operations involving matrices are accomplished by premultiplication of the matrix by certain nonsingular matrices : (a) interchanging two rows, (b) adding, to elements of one row, numbers proportional to the elements of another row, (c) multiplying elements of a row by a nonzero scalar. The same operations involving columns are performed via postmultiplication. 490. Prove that every matrix can be represented as PRQ, where P and Q are nonsingular matrices and R is a diagonal matrix of the form
/ 1
R=
0
0
*491. Prove that every matrix may be represented as a product of the matrices E+ocetk, where eais a matrix whose element of the ith row and kth column is unity, and all other elements are zero. *492. Prove that the rank of a product of two square matrices of order n is not less than r1+r2 n, where r1and r2are the ranks of the factors.
82
PART 1. PROBLEMS
493. Prove that every square matrix of rank 1 is of the form
7 A1E-Li
A2 [41 An
Al 1 12 A3 P2
Al 1 l n A2
[In
An 11 2
An 1-1'
*494. Find all third-order matrices whose squares are 0. *495. Find all third-order matrices whose squares are equal to the unit matrix. *496. Let the rectangular matrices A and B have the same number of rows. By (A, B) denote the matrix obtained by adjoining to A all the columns of B. P:ove that the rank of (A, rank of A + rank of B. *497. Prove that if A' = E, then the rank of (E+ A)+ the rank of (E A)=n, where n is the order of the matrix A. *498. Prove that the matrix A with the property A2= E can be represented in the form PBP-1, where P is a nonsingular matrix and B is a diagonal matrix, all elements of which are equal to 1. 499. Find the condition which a matrix with integral elements must satisfy so that all the elements of the inverse are also integral. 500. Prove that every nonsingular integral matrix can be represented as PR, where P is an integral unimodular matrix, and R is an integral triangular matrix all the elements of which below the principal diagonal are zero, the diagonal elements are positive, and the elements above the principal diagonal are nonnegative and less than the diagonal elements of that column. *501. Combine into a single class all integral matrices which are obtained one from the other by premultiplication by integral unimodular matrices. Compute the number of classes of nth-order matrices with a given determinant k. 502. Prove that every integral matrix can be represented as PRQ, where P and Q are integral unimodular matrices and R is an integral diagonal matrix. 503. Prove that every integral unimodular matrix of second order with determinant 1 can be represented as a product of powers (positive and negative) of the matrices A 4 0 11 l) and B= /1 0\ 1 1.
CH. 4. MATRICES
83
504. Prove that every second-order integral unimodular matrix can be represented in the form of a product of the powers of the matrices A = (1 ) /0 1 \ C= 0j 0 1 and 505. Prove that every third-order integral matrix, different from unit matrix, with positive determinant and satisfying the condition A2 =E can be represented in the form QCQ-1, where Q is an integral unimodular matrix and C is one of the matrices 1 0 0 0 0 1 0 0 1 or 0 (1 1 0 1 0 . 0 0 1
Sec. 2. Rectangular Matrices. Some Inequalities 506. Multiply the matrices: (1 (3 1 /32 2 1 and 2 and 2 ,.. 1 ) (a) (3 u 1 1 ); (b) 1;) 1 21 ) 3 1 0 (2 2) 4 . and (1 2 3); (d) (1 2 3) and (c) ( 1 1 3 507. Find the determinant of the product of the matrix 3 2 1 2) by its transpose. / \ 4 1 1 3 b, c1) 508. Multiply the matrix by its transpose and a2 b2 C2 apply the theorem on the determinant of a product. 509. Express the mth-order minor of the product of two matrices in terms of the minors of the factors. 510. Prove that all the principal (diagonal) minors of the matrix AA are nonnegative. Here, A is a real matrix, and A is the transpose of A. 511. Prove that if all the principal kth-order minors of the matrix AA are zero, then the ranks of the matrices AA and A are less than k. Here, A is a real matrix and A is its transpose. 512. Prove that the sums of all diagonal minors of a given order k computed for the matrices AA and AA are the same.
84
PART I. PROBLEMS
513. Using multiplication of rectangular matrices, prove the identity (a? + a2 + . . . + (bf + b3+ . . . + Ni) (a, b1+ a2 b2+ . . . + an b)2 =
i(k
(a, b k ak b1)2
514. Prove the identity k bi I ai bk a

k
2
Here, a1, b1are complex numbers and b; are the conjugates of b,. 515. Prove the Bunyakovsky inequality n n 2 (E a bi) a? i=1 i=i i=1
for real a1, b, by proceeding from the identity of Problem 513. 516. Prove the inequality
12
n a1 12.E
alb;
~ i
.
b 1 12
=1
i=li=t
for complex a1, bi *517. Let B and C be two real rectangular matrices such that (B, C) = A is a square matrix [the symbol (B, C) has the same meaning as in Problem 496]. Prove that I A I 2I BB I ICC I. *518. Let A = (B, C) be a rectangular matrix with real elements. Prove that I AA I 1 BB I. I CC I .
519. Let A be the rectangular real matrix

. . . a, 7an a12 a a22
a,
1
A=
a a,n2 tun
Prove that I AA
E
k=1
a?k
Eazk
k= 1
E a,k.
k=1
CH. 4. MATRICES
85
520. Let A be a rectangular matrix with complex elements and A* the transpose of the complex conjugate of A. Prove that the determinant of the matrix A* A is a nonnegative real number and that this determinant is zero if and only if the rank of A is less than the number of columns. 521. Let A=(B, C) be a complex rectangular matrix. Prove -1B*B1 C*C . that IA*A1-522. Prove that if laid 5 M, then the modulus of the determinant all a 12 a21 a22 an1 ant does not exceed Mnnn12. *523. Prove that if aik are real and lie in the interval 0 then the absolute value of the determinant made up of the
n+1
a1 a2 a,,,,
numbers aik does not exceed Mn 2 n X (n+ 1)
524. Prove that for determinants with complex elements the estimate given in Problem 522 is exact and cannot be improved. 525. Prove that for determinants with real elements the estimate given in Problem 522 is exact for n =2m. 526. Prove that the maximum of the absolute value of the determinants of order n having real elements which do not exceed 1 in absolute value is an integer divisible by 2n-1. *527. Find the maximum of the absolute value of the determinants of orders 3 and 5 made up of real numbers that do not exceed 1 in absolute value. *528. The adjoint of the matrix A is a matrix whose elements are minors of order n-1 of the original matrix in the natural order. Prove that the adjoint of the adjoint is equal to the original matrix multiplied by its determinant to the power n-2. *529. Prove that the mth-order minors of an adjoint matrix are equal to the complementary minors of the appropriate minors of the original matrix multiplied by On'-1. 530. Prove that the adjoint of a product of two matrices is equal to the product of the adjoint matrices in that order. 531. Let all combinations of numbers 1, 2, ..., n taken m at a time be labelled in some fashion.
86
PART I. PROBLEMS
Given an n x n matrix A= (aik). Let Aar, be the mth-order minor of A, the row indices of which form a combination with the index a, the column indices, a combination with the index [3. Then, using all such minors, we can construct a matrix Am = (Acco) of order C;;7'. In particular, =A, AL _ 1is the adjoint of A. Prove that (AB); = AL, BL,, E;,,= E, (A-1)L,= (A ,') -1. 532. Prove that if A is a "triangular" matrix of the form 7 an 0 a12 am \
a22 a'' -n
A=
then under an appropriate labelling of the combinations, the matrix AL, will also be triangular. 533. Prove that the determinant of the matrix A;,, is equal to A lc n 534. Let the pairs (1, k), i =1, 2, ..., n; k =1, 2, ..., m, be labelled in some fashion. The Kronecker product of two square matrices A and B of orders n and m, respectively, is the matrix C= =A x B of order nm with elements c, ,= bk, k, where a, is . Prove the index of the pair (ii, k1), a2the index of the pair (i2, that (a) (A A x B = (Ai x B) (A 2 X B), (b) A x (Bi B2) = (A x B2) (A x B2), (c) (A' x B1) (A" x B") = (A' A") x (B' B"). *535. Prove that the determinant A x B is equal to I Al"' 536. Let the matrices A and B of order mn be partitioned into n2square submatrices so that they are of the form
All
Al2 A2n
,
B11 B12
A=
A21 A22
B=
B"
\ Bn
B22
Bin \ B2r, B,,
..... n2 An1 A 2
,,
1 Bra
where .11,k and Bo, are square matrices of order m. Let their product be C and let it be partitioned in the same way into submatrices Ca,. Prove that C;lc = A a Bik + A r 2 B2k . + Ain
4. MATRICES
87
Thus, multiplication of submatrices is performed by the same formal rule as when numbers take the place of submatrices. *537. Let the matrix C of order mn be partitioned into n2 equal square submatrices. Let the matrices Au, formed from the elements of the separate submatrices commute in pairs under multiplication. Form the "determinant" E + A r, A2, ... An =B from the matrices A ik. This "determinant" is a certain matrix of order m. Prove that the determinant of the matrix C is equal to the determinant of the matrix B,
CHAPTER 5 POLYNOMIALS AND RATIONAL FUNCTIONS OF ONE VARIABLE Sec. 1. Operations on Polynomials. Taylor's Formula. Multiple Roots 538. Multiply the polynomials: (a) (2x4 +x + 1) (x2 + 1), (b) (x3+ x2x 1) (x2 2x 1). 539. Perform the division (with remainder): (a) 2x43x3 + 4x2 -5x+6 by x2 -3x+ 1, (b) x3 -3x2 x 1 by 3x22x +1. 540. Under what condition is the polynomial x3+px+q divisible by a polynomial of the form x2 +mx 1? 541. Under what condition is the polynomial x 4+px2+q divisible by a polynomial of the form x2+mx+ 1? 542. Simplify the polynomial
1 +
x(x-1)
1 2
+(
(x n+ 1) n!
543. Perform the division (with remainder): (a) 4 2x3 +4x2 -6x+8 by x-1, (b) 2x5 5x3 8x by x+3,
-
(c) 4x3 + x2 (d) x2x
by x+1+i, by x-1 +2i.
544. Using Horner's scheme, compute f (x0): (a) f(x)= x 4 3x3+ 6x2 10x +16, x0 =4, (b) f(x)= x5+ (1 +20x4 (1 +30x2+7 , x0= 2i. 545. Use the Horner scheme to expand the polynomial f(x) in powers of x xo:
(a) f(x)--=x4+2x3-3x2-4x+1,
CH. 5. POLYNOMIALS AND FUNCTIONS OF ONE VARIABLE
89
(b) f(x) = x5, (c) f(x)= x4- 8x3+24)(2-50x+90, (d) f(x)=x4 + 2ix3 - (1 + Ox2 -3x + 7 + (e) f(x)= x4 +(3 -80x3 -(21+180x2 - (33 -200x +7 +181,
x0 -= 1 x0 =2;
x0= - 1 +21.
546. Use the Horner scheme to decompose into partial fractions: (b\ x4-2x2 + 3 x3-x+ (a) (x- 2)" 1 (X +1)5 *547. Use the Homer scheme to expand in powers of x:
(a) f(x + 3) where f(x)=x 4 -x3+1, (b) (x -2)4 +4(x -2)3 +6(x -2)2 +10(x -2) +20.
548. Find the values of the polynomial f(x) and its derivatives when x=xo:
4x3+6x2 -8x+10, (a) f(x)= .x5 (b) f(x) = x 4 3ix3 -4x2 + 5ix - 1,
x0 =2, x0=1 +2i.
549. Give the multiplicity of the root:
(a) 2 for the polynomial x5 - 5x4+ 7x3 -2x2 + 4x - 8, (b) -2 for the polynomial x5 +7x4 +16x3+8x2 -16x-16.
550. Determine the coefficient a so that the polynomial x5 -ax2 -ax+1 has -1 for a root of multiplicity not lower than two. 551. Determine A and B so that the trinomial Ax 4+ Bx3+ 1 is divisible by (x -1)2. 552. Determine A and B so that the trinomial Ax"-"- + Bx" + 1 is divisible by (x - 1)2. *553. Prove that the following polynomials have 1 as a triple root:
1, (a) x24 -nx" +1+ nxn -1 (b) x2n+1 (2n +1)xn+1+ (2n + 1)x" - 1,
(c) (n -2m)xn -nxn-m + nx"' - (n -2m).
90
PART I. PROI3LEMS
554. Prove that the polynomial

x2n+1
n (n + 1) (2n + 1)xn+2(11-1)(n+2) (2n+ I)

6 2
xn+1
(n 1) (n + 2) (2n +1)
xn +
n (n + 1) (2n + 1) x n-1
6
is divisible by (x-1)5and is not divisible by (x-1)6. 1)10-1divides the polynomial *555, Prove that (X-, f(x) ao + x4 -1+ , + an if and only if a0 + an =0, a2 + ...+ ai +2a2 + +n an =0, n =0, a1 + 4a2 + + n9 a ai +2ka2 +...+nkan =0. 556. Determine the multiplicity of the root a of the polynomial
x 2 aLi(x) + f(a)l f (x) + f (a)
where f(x) is a polynomial. 557. Find the condition under which the polynomial x5 + ax3+ b has a double root different from zero. 558. Find the condition under which the polynomial x5+ 10ax3 +5bx+ c has a triple root different from zero. 559. Prove that the trinomial xn +axn-m+b cannot have nonzero roots above multiplicity two. 560. Find the condition under which the trinomial xn +axn-rn+ b has a nonzero double root. *561. Prove that the k-term polynomial
al xPz +
a2xPz + . . . + ak xpk
does not have nonzero roots above multiplicity (k 1). *562. Prove that every nonzero root of multiplicity k 1 of the polynomial a, xPz + a2 xP2 + . . . + ak xPk satisfies the equations
a1 x" (PO= a2xP2 Cp' (PO= . . . = ak xP k (Pk)
91
where cf,(t) = (t Pi) (t PP (t PO (t PO and conversely. *563. Prove that a polynomial is divisible by its derivative if and only if it is equal to a0(x x)n. 564. Prove that the polynomial
1+ 2 + I 1 2
xn n!
does not have multiple roots. 565. Prove that for x0 to be a root of multiplicity k of the numerator of the fractional rational function f(x) = 9
() ' the denominator w(x) of which does not vanish for x = x0, it is necessary and sufficient that
()
J (x0)=f' (X0) = =f (k-1)(x0)= 0, P (xo) O.

566. Prove that the fractional rational function f (x) = (x) w (x can be represented in the form
)
o)+ f(x)=f(x0)+ f' (1x) (x x
+ f(n)X) ()C 'COY'

F (x)
w (x)
x. on +1
where F(x) is a polynomial. It is assumed that w(xo) 00 (Taylor's formula for a fractional rational function). *567. Prove that if x0is a root of multiplicity k of the polynomial f1(x) f2 (x) f, (x) fi (x), then x0 is a root of multiplicity k+ 1 of the polynomial f1(x) f2 (x0)f 2 (x) (x0) if this latter polynomial is not identically zero, and conversely. *568. Prove that if f(x) does not have multiple roots, then (x)P f(x) f"(x) does not have roots of multiplicity higher than n 1, where n is the degree of f(x). *569. Construct a polynomial f(x) of degree n, for which [f' (x)]2 f(x)f"(x) has a root xoof multiplicity n 1, which is not a root of f(x).
92
PART I. PROBLEMS
Sec. 2. Proof of the Fundamental Theorem of Higher Algebra and Allied Questions 570. Define 8 so that for Ix' <8 the polynomial x5 -4x3 +2x is less than 0.1 in absolute value. 571. Define a so that 1f(x)f(2)1< 0.01 for all x satisfying the inequality Ix-21 < 8;f(x)= x43x3 +4x +5. 572. Define M so that for Ix! > M I x44x3 +4x2 +2 1>100. 573. Find x so that I f(x)I < I f(0)1 where (a) f(x)= x5 3ix3 +4, (b) f(x) = x53x3 +4. 574. Find x so that Ifix)1< lf(1 )1 where (a) f(x)=x 4 -4x3+2, 4x +5, (b) f(x)= x 4 - 4X3 6X2 -
(c) f(x)=x 4-4x+5.

575. Prove that if z i = a (1 i), 0 <a< If(z) where then
I < 1/ 5
f (z)=(1 + i) z5 +(3 5i) z4 (9 +5i) z3 7(1i)z2 +2(1+3i)z+4i.

*576. Prove that if f(z) is a polynomial different from a constant, then, in arbitrarily small neighbourhood z0, there is a z, such that I f (z,) I > I f (zo) I. 577. Prove the d'Alembert lemma for a fractional rational function. 578. Prove that the modulus of a fractional rational function reaches its greatest lower bound as the independent variable varies in a closed rectangular domain. 579. It is obvious that the theorem on the existence of a root does not hold for a fractional rational function. Thus, the function has no root. What prevents 'proving' this theorem by the scheme of that for a polynomial?
93
*580. Let f(x) be a polynomial or a fractional rational function. Prove that if a is a root off(z)f(a) of multiplicity k and f(a) 0, then for a sufficiently small p there will be, on the circle I za = P, 2k points at which I f (z) = I f (a) I. *581. Prove that if a is a root of f(z)f(a) of multiplicity k, then for a sufficiently small p there will be, on the circle I z a I= p, 2k points at which Re (f (z)) = Re (f (a)) and 2k points at which Im (z))=Im (f (a)). Here, f (z) is a polynomial or a fractional rational function. Sec. 3. Factorization into Linear Factors. Factorization into Irreducible Factors in the Field of Reals. Relationships Between Coefficients and Roots 582. Factor the following polynomials into linear factors: (a) x3 -6x2 + 1 lx 6, (b) x4 +4, (c) x4 + 4x3 + 4x2 + 1, (d) x4 10x2 + 1. *583. Factor the following polynomials into linear factors: (a) cos (n arc cos x), (b) (x+ cos 0+ i sin 0)"+ (x+ cos 0 i sin 0)n, (c) xm xm-i+ Xm-2 . . . ( 584. Factor the following polynomials into irreducible real factors : (a) x4 +4, (b) x6 +27, (c) x4 + 4x3 +4x2 + 1, (d) x'n an +2, (e) x4 ax2 +1, 2 < a < 2, (f) X2n X n +1 585. Construct polynomials of lowest degree using the following roots: (a) double root 1, simple roots 2, 3, and 1 + (b) triple root 1, simple roots 3 and 4, (c) double root i, simple root 1 i. 586. Find a polynomial of lowest degree whose roots are all roots of unity, the degrees of which do not exceed n.
94
PART I. PROBLEMS
587. Construct a polynomial of lowest degree with real coefficients, using the roots: (a) double root 1, simple roots 2, 3 and 1 +i, (b) triple root 2 3i, (c) double root i, simple root 1 i. 588. Find the greatest common divisor of the polynomials: (a) (x 1)3 (x +2)2 (x 3) (x 4) and (x 1)2 (x +2) (x +5), (b) (x I) (x2 1) (x3 -1) (x4 -1) and (x + 1) (x2 +1) (x3 + 1). (x4 +1), (c) (x3 -1) (x2 -2x+ 1) and (x2 1)3. *589. Find the greatest common divisor of the polynomials xm 1 and xn-1. 590. Find the greatest common divisor of the polynomials
xm + a"' and xn+an.

591. Find the greatest common divisor of the polynomial and its derivative:
(a) f(x)=(x 1)3(x + 1)2(x 3), (b) f(x)=(x 1) (x2-1) (x3-1) (x4 -1), (c) f(x)= xin+n Xn1Xn +1.
592. The polynomial f(x) has no multiple roots. Prove that if x0is a root of multiplicity k > 1 of the equation f =0, then the equation f( v u 0 has xo as a root of multiplicity k 1. It is assumed that v (xo) 00, v' (x0)00. 593. Prove that X2 X + 1 divides x3'n +.x 3r: +1+x3P+2. 594. When is x3"' x3n +1+x3P+ 2 divisible by x2 x + 1? 595. What condition is necessary for x4 +x2 + 1 to divide x3"' + x3,z+1+ x3p +2? 596. What condition is necessary for x2+x+ 1 to divide x2"i+
,1 x )=
:((.2 x
+x'n+1? 597. Prove that

x kat
xlca2-1-1 ,k ak +k 1 A
is divisible by xk-1+xk-2+...+1.
95
598. For what values of m does x2+x+ 1 divide (x +1)"' 1? 599. For what values of m does x2+x+ I divide (x + 1)"' + +xnz+ 1 ? 600. For what values of m does (x2+x+ 1)2 divide (x + xn21? 601. For what values of m does (x2+x+ 1)2 divide (x+ 1)"'+ + xfr n + 1? 602. Can (x2+x +1)3divide the polynomials (x +1)rn + x"i + I and (x +1)m xln 1? 603. Transform the polynomial
x 1 x (x-1) 1 2 . . . (xn +l) _ on x(x-1) 1.2...
by assigning to x the values 1, 2, ..., n in succession. (Compare with Problem 542.) 604. For what values of m does X(x) divide X(xn2)? (X is a cyclotomic polynomial.) Prove the following theorems: 605. If f(x") is divisible by x 1, then it is also divisible by xn 1. 606. If f(x") is divisible by (xa)k, then it is also divisible by (x"an)k for a O. 607. If F (x) f, (x3) + xf, (x3) is divisible by x' + x + 1 , then f, (x) and f 2 (x) are divisible by x-1. *068. If the polynomial f(x) with real coefficients satisfies -=[soi(x)]2 + the inequality f(x) 0 for all real values of x, then f(x)+ FP2(x)12 where cp,(x) and c, o 2(x) are polynomials with real coefficients. 609. The polynomial f(x) aoxn + + . . . + a has the roots x1, . . x. What roots do the following polynomials have: (a) c/oxn a, .xn + a2 (b) an xn + a_, 1 . . . + ( 1)11a , . . . a0 , f(n) (a)
n!
(c) f (a) + f (a) X+ f' (a) A2 + 12 1 2 b2 xn-2+ (d) xn + a, bxn-1 + a
+an t,n?
610. Find a relationship between the coefficients of the cubic equation .x 3+px2+ qx+r =0 under which one root is equal to the sum of the other two.
96
PART I. PROBLEMS
611. Verify that one of the roots of the equation 36x3 -12x2 - 5x+ 1 =0 is equal to the sum of the other two, and solve the equation. 612. Find a relationship among the coefficients of the quartic equation x4 + ax3+ bx2+ cx+ d= 0 under which the sum of two roots is equal to the sum of the other two. 613. Prove that the equation which satisfies the condition of Problem 612 can be reduced to a biquadratic equation by the substitution x=y+a for an appropriate choice of a. 614. Find a relationship among the coefficients of the quartic equation x4 + ax3+ bx2+ cx+ d= 0 under which the product of two roots is equal to the product of the other two. 615. Prove that the equation which satisfies the hypothesis of Problem 614 may be solved by dividing by x2 and substituting y=x + ax- (for a 616. Using Problems 612 to 615, solve the following equations:
(a) x4 -4x3 + 5x2 2x 6 =0, (b) x4 +2x3 + 2x2+10x +25 =0, (c) x4 +2x3 + 3x2 + 2x 3=0, (d) x4 + x3 -10x2 2x+ 4=0.
617. Define A so that one of the roots of the equation x3 7x + + A=0 is equal to twice the other root. 618. Define a, b, c so that they are roots of the equation
x3 ax2 + bx c= 0.
619. Define a, b, c so that they are roots of the equation
x3+ax2+bx+ c= 0.
620. The sum of two roots of the equation
2x3 x2 -7x+A= 0 is equal to 1. Determine A.

621. Determine the relationship between the coefficients of
the equation x3+px+q=0 under which x3= 1 + 1 . Xy )C2 622. Find the sum of the squares of the roots of the polynomial
xn+ai xn-l+ . + an.
97
*623. Solve the equation xn aix n--1 a2 x-2+
+ an = 0
knowing the coefficients al and a2 and that its roots form an arithmetic progression. 624. Do the roots of the equations (a) 8x3-12x2 -2x + 3 = 0, (b) 2x4 + 8x3 +7x2 2x-2=0 form arithmetic progressions? 625. Given the curve
y=x4+ax3+bx2+cx+d.
Find a straight line whose points M1, M2, M3, M4of intersection with the curve intercept three equal segments M1M2=M2M3==M3.3/4. Under what condition does this problem have a solution? *626. Form a quartic equation whose roots are a,, a, 1 *627. Form a sextic equation with the roots 1 1 1 1 a,
, 1cc 1 cc ' 1 o: ' cc '
1 1 -cc
628. Let f (x)=(x (x x 2) ... (x x). Find f' (xi), f (xi) and prove that af ' (xi) 1 axi = 2 f (xi) 629. Prove that if f (x1)- f"(x1)=0 but f' (x1) 0 0, then
1
1=2
xlXi
_O.
630. The roots of the polynomial xn+a1xn-4+ ...+a form an arithmetic progression. Determine f' (xi). Sec. 4. Euclid's Algorithm 631. Determine the greatest common divisor of the following polynomials: (a) x4 + x3-3x2 -4x-1 and x3 +x2 x-1; (b) x5+ x4x3-2x 1 and 3x4 +2x3 +x2 +2x-2;
4. 1215
98
PART I. PROBLEMS
(c) x6 7x4 + 8x8 7x +7 and 3x5-7x3 +3x2 -7; (d) x5-2x4 +x3+ 7x2 12x + 10 and 3x4 6x3+ 5x2+ 2x 2; (e) x6 + 2x4 4x3 3x2 + 8x 5 and x' + x2 x + 1 ; (f) x5+ 3x4 12)0 52x2 52x 12 and x4 + 3x3 6x2 22x 12 ; (g) x' + x4 x' 3x2 3x 1 and x4 2x' x2 2x + 1; 6x2 + 41/ 2 x + 1 ; (h) x4 10x2 + 1 and x4 -4y'2 (i) x4 + 7x3 + 19x2 +23x + 10 and x4 +7x3+ 18x2 +22x+ 12; (j) x4 -4x3 + 1 and x3-3x2+ 1 ; (k) 2x6 5x' 14x4 + 36x' + 86x2 + 12x 31 and 2x5 -9x4 + 2x3+ 37x2 + 10x 14 ; (1) 3x6 x5 9x4 14)0 11x2 3x 1 and 3x5+8x4 +9x3+15x2 +10x+9.
632. Using Euclid's algorithm, choose polynomials M1(x) and M2 (x) so that fl(x) M2 (x) +12 (x) Mi.(x) = (x) where (x)
is the greatest common divisor off,. (x) and 1 2 (x): (a) fl(x)=x4+2x3x2 4x -- 2, 12 (x) =x 4+ xs x2-2x 2 ; (b) fl (x)=x5 +3x4 +x3 +x2 +3x+ 1, f2(x)=x4 +2x3+x+2; (c) f, (x)=x6 -4x5+11x4-27 x3+37x2 35x + 35, f2 (x)=x5 3x4 +7x3-20x2+ 10x 25 ; (d) fl (x)=3x7 + 6x6 -3x5+ 4x4 + 14x3 6x2 4x + 4, 12 (x)=3x6 -3x4+ 7x3 6x + 2 ; (e) jel(x)=3x5+5x4 16x3 6x2 5x 6, f2 (x)-3x 4 4x3 x2 x-2; (f) fl(x)=4x4 -2x3 16x2+ 5x + 9, 12 (xl =2.x3x2 5x + 4.
99
633. Using Euclid's algorithm, choose polynomials M1(x) and M2 (x) so that ft (x) M2 (x)+f2 (x) M1(x)=1 : (a) fi(x)=3x3 2x2 +x + 2, f2 (x) = X2 X+ 1 ; 1; (b) (x) = x4 x3 4x2+ 4x + 1, f2 (x) = X2 (c) f1 (X) = X5 5x 4 2X3+ 12X2 2x + 12, f2 (x)=x3 5x2 3x+ 17; (d) f1 (x)=2x4+3x3 3x2 5x + 2, f2 (x)=2x3+ x2 x 1; (e) f1(x)=3x4 5x3+ 4x2 2x + 1, f2 (x)=3x3 2x2+x 1 ; (f) f1(x)=x5+5x4+9x3+7 x2+5x +3, f2 (x)=x4 +2x3+2x2 +x+ 1. 634. Use the method of undetermined coefficients to choose M]. (x) and M2 (x) so that (x) M2 (x)+f2(x) M1(x) = 1 : (a) ( fi x)= x4 4x3+ 1, f2 (x) = x3- 3x2 + ; f2 (x)=(1 x)2 ; (b) f1(x)=x 3, ( (1 x)4. f2(x)= (c) f1(x)= x4, 635. Choose polynomials of lowest degree, M1(x), M2 (x), so that (a) (x4 2x3-4x2 + 6x + 1) M1(x) + (x3 5x 3) M2 (X) = X4 ; (b) (x4+ 2x3+x + 1) M1(x) + (x4+ x32x2 +2x 1) M2 (X) = X3 2x. 636. Determine the polynomial of lowest degree that yields a remainder of: (a) 2x when divided by (x 1)2and 3x when divided by (x-2)3; (b) x2+ x + 1 when divided by x42x3 2x2+ 10x 7 and 2x2 3 when divided by x4 2x3 3x2+ 13x 10. *637. Find polynomials M (x) and N (x) such that M (x)+ (1 x)^ N (x)=1. 638. Let f1(x) M (x) +f2(x) N (x) = a (x) where a (x) is the greatest common divisor of f1(x) and f2 (x). What is the greatest common divisor of M (x) and N (x)?
4"
100
PART I. PROBLEMS
639. Separate the multiple factors of the polynomials:
(a) x5 6x4 4x3+ 9x2 + 12x +4, (b) x5 10x3 20x2 15x 4, (c) x6 15x4 + 8x3+ 51x2 72x + 27, (d) x5 6x4 +16x3 24x2 + 20x 8, (e) x6 2x5 x 4 2X3 +5X2 + 4x +4, (f) x7 3x6+ 5x57 x 4 +7x2 5x2 +3x-1, (g) x8 + 2x7 +5x6 +6x5 +8x4 +6x3+5x2 +2x+ 1.
Sec. 5. The Interpolation Problem and Fractional Rational Functions 640. Use Newton's method to construct a polynomial of lowest degree by means of the given table of values: 1,, x I 0 1 23 4 ib \ X I 1 01 23
"/ f (x) I 1 2 3 4 6' 9 25 X 1 4 4 4 (c) 3 5 2 f(x) 1 2 2
f(x) I
65032
'
I 123 find f (2); (d) X f(x) I 5 6 1 4 10
641. Use Lagrange's formula to construct a polynomial by the given table of values :
x1 1 2 3 4
'a' y I 2 1 4 3 '
\ X Ili (b ) y I 1 2
i
3 4
*642. Find f (x) from the following table of values:

X
I1e l E2
Eni
f (x)
1 2 3 . . n
Ek = cos
2rck
+ i sin
2rck
n -
643. A polynomial f (x), whose degree does not exceed n-1, takes on values yl, Y25 , yrz in the nth roots of unity. Find f (0). *644. Prove the following theorem: so that f(x)= 1[f(xi)+f(x2)+ +f(x.)1
for any polynomial f (x) whose degree does not exceed n 1 it is necessary and sufficient that the points x1, x2, ..., x,, be located on n circle with centre at xoand that they divide it into equal parts.
101
*645. Prove that if the roots x1, x 2, ..., xn of a polynomial

cp (x) are all distinct, then
cp' (xi)
i =1
0 for 10..s<n-2.
n-1
646. Find the sum E

=
c?' (xi)
(notations are the same as in
Problem 645). 647. Derive the Lagrange interpolation formula by solving the following system of equations: a, + x1+ . . . + an _ix7-1= yi, no + a, x2+ . . . an_ix2-1=Y2,
no+ xn + . . . + an_ixn n-l=yn.

*648. Use the following table of values to construct a polynomial of lowest degree :
xI012... n Y 1 1 2 4 . . . 2n'
*649. Use the following table of values to construct a polynomial of lowest degree: x1 0 1 2 . . .n
yllaa2 ...an'
*650. Find a polynomial of degree 2n which upon division by x (x 2)...(x 2n) yields a remainder of 1, and upon division by (x-1) (x-3)...[x (2n 1)] yields a remainder of 1. *651. Construct a polynomial of lowest degree, using the table of values
x1 1 2 3 . . .n 1 1 1 n 23
*652. Find a polynomial of degree not exceeding n 1 that
satisfies the condition f (x)=x a at the points x1, x2, ..., xn, x1 0 a, i= 1, 2, ..., n. *653. Prove that a polynomial of degree k c which assumes integral values for n+ 1 successive integral values of the independent variable, takes on integral values for all integral values of the independent variable.
102
PART I. PROBLEMS
*654. Prove that a polynomial of degree n, which takes on integral values for x =0, 1, 4, 9, ..., n2, assumes integral values for all squares of the natural numbers. *655. Decompose into partial fractions of the first type :
x2
(a)(x 1) (x+ 2) (x+ 3) '
1
(b)(x 1) (x 2) (x 3) (x 4) '
(x-1) (x2+ 1) ' 1

x4 +4 '
3 +x
(d)
1
x2 , x 1 ,
1
, ) (h
(e)
x2 -1
(g) xn 1 n1
xn+ 1
x (x 1) (x 2) . . . (x n) ' (2n)1
x (x2 1) (x2 4) . . . (x2 n2) '
(k)
cos (n arc cos x)
*656. Decompose into real partial fractions of the first and second types:
(a) x81
; (b) x4 16 ; (c) x4 4 ;(d) x8 +27

m < 2n + 1; m<2n+ 1;
x2
x2
(e) x2n x rn +i_i

m
(f) 2/14-1+ 1 9 x
(g)
(i)
xsnl_ 1 ; (h))0xna: , m < n ;

1
x (x2 + 1) (x2+ 4) . . . (x2 + n2)
*657. Decompose into partial fractions of the first type: (a) (x2 1)" (d)
(xn 1)2 ;
1
( b)
(x 2 1)2
1
(c) ' (x 1)3(x + 1)2(x 2) '

1
5x2 + 6x 23
(e) xm(1 x)n ; (h) [f (42

g (x)
(f) (x 2 a2)n '
a 0;
(g) (xa a2y:
where f (x x1) (x x 2) . . . (x xn) is a polynomial with no multiple roots and g (x) is a polynomial of degree less than 2n.
103
658. Decompose into real partial fractions of the first and second type: (a) (x +1) (x2+ 1)2 , (b) x (x+ 1)2 (x2 + x + 1)2 '
(c)
1
(X4 - 1 )2
2x- 1
(d)
1 (x 21/
1)2
659. Let cp (x)= (x x1) (x x 2) ... (x x ). Express the following sums in terms of cp (x): (a)
xxi ;
(b)
E x_xi x'
(c)
(x_x02
*660. Compute the following sums, knowing that x1, x2, ... are roots of the polynomial co (x): 1 1 1 cp (x) = x3 3x 1; (a) 2x1 2_ x2 + 2x2 ' (b)
1 3x, + 2 + 1 1 3x2+ 2 + 4 3x2 + 2 '
cp(x)=x3 +x2 -4x+ 1; (c) xi-2x1+1 1 1

4 2x2 + 1
1
4 2x2 + 1 '
cp (x)= x3+ x2 1 661. Determine the first-degree polynomial which approximately assumes the following table of values: x1 0 1 2 3 4
y I 2.1 2.5 3.0 3.6 4.1
so that the sum of the squares of the errors is a minimum. 662. Determine the second-degree polynomial which approximately assumes the table of values x1 0 1 2 3 4
Y I 1 1.4 2 2.7 3.6
so that the sum of the squares of the errors is a minimum.
Sec. 6. Rational Roots of Polynomials. Reducibility and Irreducibility over the Field of Rationals
663. Prove that if 11is a simplified rational fraction that is a root of the polynomial f (x) aox" + + . . . +a" with integral coefficients, then
104
PART 1. PROBLEMS
(1) q is a divisor of a0, (2) p is a divisor of a, (3) pmq is a divisor of f (m) for any integral m. In particular, pq is a divisor of f (1), p+q is a divisor of f (-1). 664. Find the rational roots of the following polynomials: (a) x3 -6x2 +15x 14, (b) X 4 2 X3- 8 X2 +13x-24, (c) x5-7x3 -12x2 +6x+36, (d) 6x4 +19x3-7x2 -26x +12, (e) 24x4 -42x3-77x2 +56x +60, (f) x6 -2x4 -4x3+4x25x +6, (g) 24x6 +10x4 x3-19x2 5x + 6, (h) 10x4 -13x3 +15x2 -18x 24, (i) x4 +2x3-13x2 -38x-24, (j) 2x3+3x2 +6x 4, (k) 4x4 -7x2 -5x-1, (1) x4 +4x3-2x2 -12x+9, (m) x6 + x4 6x3 14)0 llx 3, (n) x6 6)0+ 11x4 x3 -18x2 + 20x 8. *665. Prove that a polynomial f (x) with integral coefficients has no integral roots if f(0) and f (1) are odd numbers. *666. Prove that if a polynomial with integral coefficients assumes the values 1 for two integral values x1and x2 of the independent variable, then it has no rational roots if 1x1 x2 1> 2. However, if 1 x1 x2I 2, then the only possible rational root 1 is (x1+x2). *667. Use the Eisenstein criterion to prove the irreducibility of the polynomials (a) x4 8x3+12x2 6x+2, (b) x5 12x3 +36x-12, (c) X 4 - X3 2x +1. *668. Prove the irreducibility of the polynomial i X p (x) = xP p prime. x -1 *669. Prove the irreducibility of the polynomial
X IC(x)_
pk
Xk_i
p prime.
105
*670. Prove that the polynomial f(x)=a,xn+aixn-i+ +a with integral coefficients that has no rational roots is irreducible if there exists a prime number p such that a, is not divisible by p; a2, a3, ..., a, are divisible by p and anis not divisible by p2. *671. Let f(x) be a polynomial with integral coefficients for which there is a prime number p such that a, is not divisible by p; ak+i, ak,, ...,a are divisible by p and an is not divisible by p2. Prove that in that case f(x) has an irreducible factor of degree ?.-nk. 672. Using the method of factorization into factors of the values of a polynomial with integral values of the variable, decompose the following polynomials into factors or prove their irreducibility: (a) x 43x2+1, (b) x4 + 5x3 3x2 5x +1, (c) x4 + 3x3 2x2 2x +1,
(d) x 4 -X3- 3X2 +2x + 2.

673. Prove that a polynomial of degree three is irreducible if it has no rational roots. 674. Prove that the fourth-degree polynomial x4 + ax3+bx2+ +cx+d with integral coefficients is irreducible if it has no integral roots and is not divisible by any polynomial of the form
x2 +
cm amt
a' m2
X -F
where m are divisors of the number d. Polynomials with fractional coefficients may be disregarded. Polynomials like those of Problems 614, 615 are a possible exception. 675. Prove that the fifth-degree polynomial x5+ax4+bx3+ +cx2+dx+e with integral coefficients is irreducible if it has no integral roots and is not divisible by any polynomials with integral coefficients of the following form:
X2 + am' cm2 dn+ be m3 72 2+ ae dm x m
where m is a divisor of e, [17= m.
676. Factor the following polynomials and prove their irreducibility using Problems 674, 675: (a) x4 3x3+ 2x2 + 3x 9, (b) x4 3x3+ 2x2+ 2x 6,
106
PART I. PROBLEMS
(c) x4 + 4x3 6x2 23x 12,
(d) x' + x4 4x3 + 9x2 6x + 6.

677. Find the necessary and sufficient conditions for reducibility of the polynomial x4+px2+q with rational (possibly fractional) coefficients. 678. Prove that, for the reducibility of a fourth-degree polynomial without rational roots, it is necessary (but not sufficient) that there exists a rational root of a cubic equation obtained in a solution by the Ferrari method. *679. Prove the irreducibility of the polynomial f (x) = (x;)-1; a1, a2, ..., an are distinct in=(xa1)(xa2) tegers. *680. Prove the irreducibility of the polynomial f (x)= (x a,) +1 for distinct integers a1, a 2, a =(xa1)(xa2) with the exception of
(x a) (x a -- 1) (x a-2) (x a 3)+1 = [(x a-1) (x a 2)-112 and (x a) (x a 2)+ I =(x a 1)2. *681. Prove that if an nth-degree polynomial with integral coefficients assumes the values 1 for more than 2m integral values of the variable (n=2m or 2m +1), then it is irreducible. *682. Prove the irreducibility of the polynomial f (x)=(x a1)2(x a 2)2 . . . (x a)2+1 if a1, a2, a are distinct integers.
*683. Prove that the polynomial f (x) with integral coefficients which takes on the value +1 for more than three integral values of the independent variable cannot assume the value 1 for integral values of the independent variable. *684. Prove that an nth-degree polynomial with integral coefficients assuming the values 1 for more than integral values of the independent variable is irreducible for n 12. *685. Prove that if a polynomial ax2+bx+ 1 with integral coefficients is irreducible, then so is the polynomial a [cp (x)]2+ + b (x)+1 where cp (x)=(x al) (x a2) ... (x an) for n 7.
Here, al,
anare distinct integers.
107
Sec. 7. Bounds of the Roots of a Polynomial 686. Prove that the roots of the polynomial aox" +aixn -1+ + ... +anwith real or complex coefficients do not exceed, in absolute value,
(a) 1 +max (b) p +max
ak
ao
, k = 1, 2, . . . , n;
1
ak ao P k
k=1, 2, ..., n; p is any positive number;

k
ak (c) 2 max VI ao , k = 1, 2, . . k -1
n;
(d)
al
a0
ak +max 1/' a1 , k = 1, 2, .. . , n.
687. Prove that the moduli of the roots of the polynomial aoxn+aixn-1+ +a do not exceed a unique positive root of the equation boxnbixn-1b2xn-2 ... be, where 0< <bo < I a1, al I, b2 ?- I a21, b an !. 688. Prove that the moduli of the roots of the polynomial f (x)= aoxn + ar xn-r + +an, ar 0, do not exceed
r
(a) 1 + 1/max I `` 1 , k =r, . . . , n; a0

(b) p+ max i ao pk -r
ak
k = r,
(c)
a,
n, and p is any positive number;

k-r
ao
+ max -V
ak ar
, k=r,
n.
689. Prove that the real roots of a polynomial with real coefficients do not exceed a unique nonnegative root of a polynomial obtainable from the given polynomial by deleting all terms (except the highest-degree one), the coefficients of which are of sign that coincides with the sign of the leading coefficient. Prove the following theorems:
108
PART I. PROBLEMS
690. The real roots of the polynomial aoxn + aixn --1 + + an with real coefficients (for ao> 0) do not exceed ak (a) 1 + 1/ max .` ao r is the index of the first negative
coefficient and ak are negative coefficients of the polynomial;

a:_ - where r is the index of the first (b) p+ Vmax as p r negative coefficient, ak are negative coefficients, and p is any positive number;
(c) 2 max V I a ak I ,ak are negative coefficients of the polynomial;

kr ak , r is the index of the first neI ar I + max 1/ a, a gative coefficient, and akare negative coefficients. 691. If all the coefficients of the polynomial f (x) are nonnegative, then the polynomial does not have positive roots. 692. If f (a) > 0, f' (a) 0, ..., f (n) (a) 0, then all real roots of the polynomial do not exceed a. 693. Indicate the upper and lower bounds of the real roots of the polynomials:
(d)
(a) x4 4x3+ 7x2 8x + 3, (b) x5 + 7x3 3, (c) x7 108x5 445x3+ 900)0+ 801, (d) x4 + 4x3 8x2 -10x + 14. Sec. 8. Sturm's Theorem 694. Form the Sturm polynomials and isolate the roots of the polynomials: (a) x3-3x 1, (b) x3+x2 -2x 1, (c) x37x +7, (d) x3 x + 5, (e) x3+3x 5. 695. Form the Sturm polynomials and isolate the roots of the polynomials: (a) x4-12x2 16x 4, (b) x4 x 1,
109
(c) 2x4 -8x3+8x2 -1, (d) x4 +x2 -1, (e) x4 + 4x3 12x + 9. 696. Form the Sturm polynomials and isolate the roots of the following polynomials: (a) x4 -2x3 -4x2 +5x+5, (b) x4 -2x3 +x2 -2x+1, (c) x4 2x3 3x2 + 2x + 1, (d) x4 x3+ x2 x 1, (e) x4 -4x3-4x2 +4x+1. 697. Form a Sturm sequence and isolate the roots of the following polynomials: (a) x4 2x3 7x2 + 8x + 1, (b) x4 4x2 +x + 1, (d) x4 -4x3 +8x2 -12x+8, (c) x4 x3 x2 x + 1, (e) x4x3 2x + 1. 698. Form a Sturm sequence and isolate the roots of the following polynomials: (b) 4x4 -12x2 +8x-1, (a) x4 -6x2-4x+2, (c) 3x4 +12x3 +9x2 -1, (d) x4 x3-4x2 +4x+1, (e) 9x4 -126x2 -252x-140.
699. Form a Sturm sequence and isolate the roots of the following polynomials:
(a) 2x5-10x3 +10x-3, (b) x6 -3x6 -3x4 +11x3-3x2 -3x+1, (c) x6 +x4 -4x3-3x2 +3x+1, (d) x6 -5x3-10x2 +2.
700. Form a Sturm sequence using the permission to divide the Sturm functions by positive quantities, and isolate the roots of the following polynomials:
(a) x4 +4x2 -1, (b) x4 -2x3 +3x2 -9x+1, (c) x4 -2x3+2x2 -6x+1, (d) x6 +5x4 +10x2 -5x-3. 701. Use Sturm's theorem to determine the number of real roots of the equation x3+ px + q= 0 for p and q real. *702. Determine the number of real roots of the equation xn +px+ q= 0.
110
PART I. PROBLEMS
703. Determine the number of real roots of the equation x5 -5ax3+5a2x+2b= 0. 704. Prove that if a Sturm sequence contains polynomials of all degrees from zero to n, then the number of variations of sign in the sequence of leading coefficients of Sturm polynomials is equal to the number of pairs of conjugate complex roots of the original polynomial. 705. Prove that if the polynomials f (x), f1(x), f2(x), . . ., fk (x) have the following properties: (1) f (x) fi (x) changes sign from plus to minus when passing through the root f (x); (2) two adjacent polynomials do not vanish simultaneously; (3) if fa (xo)=0, then f;,_, (x0) and fx4.1(x0) have opposite signs ; (4) the last polynomial f, (x) does not change sign in the interval (a, b), then the number of roots of the polynomial f (x) in the interval (a, b) is equal to the increment in the number of variations of sign in the sequence of values of the polynomials f, f1, ..., fk when going from a to b. 706. Let x, be a real root of f' (x): 1xo f' (x); A (x) ---x
A (x) is the remainder, after division of f (x) by f1(x), taken with

reversed sign; f 3 (x) is the remainder, left after dividing f1(x) by f 2(x), taken with reversed sign, and so on. It is assumed that f (x) has no multiple roots. Relate the number of real roots of f (x) to the number of variations in sign in the sequence of values of the polynomials constructed for x= co, x=xo, and x= + co. *707. Construct a Sturm sequence for the Hermite polynomials x2
Pn (x)= (
1)n e' dn edx:
and determine the number of real roots. *708. Determine the number of real roots of the Laguerre polynomials -x xn) i)n ex dn (edxn
. 1 ' , (x) = ( _
111
Determine the number of real roots of the following polynomials: *709. En(x)= 1 + x -F
X2 j_ 1 2 -I-
Xn n!
x) cin+1 (e1
*710. P(x)= (- 1)"+1 x 2n+2 e X
dxn+1
x2 1 +
n *711. P(x)= (-1211)n ( x2 + 0,7+1 d dxn
*712. P(x)=(- 1)n (X2 1)n + 2 dx"
+1dn
V x2 +1
*713. Let f (x) be a third-degree polynomial without multiple roots. Show that the polynomial F (x) = 2f (x) f" (x) - f' (x)]2 has two and only two real roots. Investigate the case when f (x) has a double or a triple root. 714. Prove that if all roots of the polynomial f (x) are real and distinct, then all roots of each of the polynomials of the Sturm sequence constructed by the Euclidean algorithm are real and distinct. Sec. 9. Theorems on the Distribution of Roots of a Polynomial Prove the following theorems: dn (.0 1)n 715. All roots of the Legendre polynomial Pn (x)dxn are real, distinct and are in the interval (- 1, + 1). 716. If all the roots of the polynomial f (x) are real, then all the roots of the polynomial a f (x)+ f ' (x) are real for arbitrary real A. *717. If all the roots of the polynomial f (x) are real and all the roots of the polynomial g (x)= aoxn + aixn + . . . + a are real, then all the roots of the polynomial
F (x)= a, f (x)+ air (x)+ ... +a f (n) (X)

are real. *718. If all the roots of the polynomial f (x)= aoxn + aixn -1+ + ... +a are real, then all the roots of the polynomial xn + mxn-1 a2 m (m I) xn-2
-
+ an m (m - 1) . . . (m -n + 1) + are real for arbitrary positive integral m.
112
PART I. PROBLEMS
*719. If all the roots of the polynomial f (x)= aoxn + aixn-1+ + ; are real, then all the roots of the polynomial + G(x)=a0 xn+C;i ai xn-l+C,a2 x4-2+...+; are real. 720. Prove that all the roots of the following polynomial are real:
xn
n\2 xn_i )
n(n 2 x n-2
+1.
*721. Determine the number of real roots of the polynomial 1. nxn xn-i_ xn 2 722. Determine the number of real roots of the polynomial
x 2,2,1-1 + x212,1-1
x2n, - F1 +
a.
723. Determine the number of real roots of the polynomial f (x)=(x a) (x b) (x c) A2(x a) B2(x b) C2(x c) for a, b, c, A, B, C. 724. Prove that
Al (p (x)= x + 2 +... + n +B xan xa2 does not have imaginary roots for real al, a2, Al, A 2, A, B. Prove the following theorems: 725. If the polynomial f (x) has real and distinct roots, then (x)]2f (x) f" (x) does not have real roots. 726. If the roots of the polynomials f (x) and cp (x) are all real, prime and can be separated, that is, between any two roots of f (x) there is a root of cp (x) and between any two roots of cp (x) there is a root off (x), then all roots of the equation of (x) + p.c,o (x)= =0 are real for arbitrary real A and *727. If all the roots of the polynomials F (x) = X f (x) + cp (x) are real for arbitrary real A and then the roots of the polynomials f (x) and cp (x) can be separated. *728. If all the roots of f' (x) are real and distinct and f (x) does not have multiple roots, then the number of real roots of the polynomial [f' (x)]2f (x) f" (x) is equal to the number of imaginary roots of the polynomial f (x). *729. If the roots of the polynomials f1(x) and f2 (x) are all real and separable, then the roots of their derivatives can be separated. A2
A2
CH. 5. POLYNOMIALS AND FUNCTIONS OF ONE
VARIABLE
113
*730. If all the roots of the polynomial f (x) are real, then all the roots of the polynomial F (x)=yf (x)+ (X+ x) f'(x) are real for y > 0 or y< n and for arbitrary real A as well. *731. If the polynomial
f (x) a0+ aix + . . . + anxn

has only real roots, and the polynomial cp (x)= bo+ bix +
+bkxk
has real roots that do not lie in the interval (0, n), then all the roots of the polynomial ao cp (0)+aicp (1) x+a2 cp (2) x2 + ... +ancp (n)xn are real. *732. If all the roots of the polynomial f(x) = a0+ + . + anxn are real, then all the roots of the polynomial ao+ aiyx + a2y (y 1)x2 + + a y (y 1) ...(y n +1) xn are real for y> n 1. *733. If all the roots of the polynomial f (x)= a0+ + . . . + axn are real, then also real are the roots of the polynomial
ao + a
x+
Y (Y -1)
2 + ... + (a+1) a2 x
y (y-1) . . (yn+ 1)
(oc+1)
(a+n-1)
xn +...
for y>n 1, a >0. *734. If all the roots of the polynomial f (x)= ao+ . . . + anxn are real, then all the roots of the polynomial ao + alwx+ a2 w4 x2+ . . . +
xn
are real for 0 < w *735. If all the roots of the polynomial a0+ai x+a2 x2+ + . . . anxnare real and of the same sign, then all the roots of the polynomial a0cos cp+alcos (cp +0) x + a2cos (cp +20) x2 + + an cos (cp +n0) xn are real. *736. If all the roots of the polynomial (ao +ibo)+(al + x + . . . +(an+ ib)xn lie in the upper half-plane, then all the roots of the polynomial + b,,xn ao+ + . . . + anxn and bo +bix+ are real and separable (the numbers a0, al, . . a., b0, b1 . . bn are real). *737. If all the roots of the polynomials cp (x) and cP (x) are real and separable, then the imaginary parts of the roots cp (x)+ + i (x) have the same signs.
,
114
PART I. PROBLEMS
*738. If all the roots of the polynomial f (x) lie in the upper half-plane, then all the roots of its derivative likewise lie in the upper half-plane. *739. If all the roots of the polynomial f (x) are located in some half-plane, then all the roots of the derivative are located in the same half-plane. *740. The roots of the derivative of the polynomial f (x) lie within an arbitrary convex contour which contains all the roots of the polynomial f (x). *741. If f (x) is a polynomial of degree n with real roots, then all the roots of the equation [f (x)]2+ k2+ rrom2=0 have an imaginary part less than kn in absolute value. 742. If all the roots of the polynomials f (x) a and f (x) b are real, then all the roots of the polynomial f (x) A are real if X lies between a and b. *743. For the real parts of all the roots of the polynomial x"+ + aix"-1+ . . . + ; with real coefficients to be of the same sign, it is necessary and sufficient that the roots of the polynomials
x" a2 x"-2 + a4 Xn-4- .

and
xn-1
. .
a3 x"-3
be real and separable. *744. Find the necessary and sufficient conditions for the real parts of all the roots of the equation x3 + ax2+ bx + c=0 with real coefficients to be negative. *745. Find the necessary and sufficient conditions for the negativity of the real parts of all the roots of the equation x4 + axs + + bx2+ cx + d= 0 with real coefficients. *746. Find the necessary and sufficient conditions for all the roots of the equation x3 + ax2+ bx + c = 0 with real coefficients not to exceed unity in absolute value.
n > 0, then all the roots *747. Prove that if a, a1?. a2 ... a of the polynomial f (x)-- aox" + aix"-1+ . + a do not exceed unity in absolute value.
115
Sec. 10. Approximating Roots of a Polynomial 748. Compute to within 0.0001 the root of the equation x3-3x2 -13x-7=0 which lies in the interval (-1, 0). 749. Compute the real root of the equation x' 2x 5 = 0 with an accuracy of 0.000001. 750. Compute the real roots of the following equations to within 0.0001: (b) x3+2x-30=0, (a) x3 -10x-5=0, (d) x3-3x2 x+2=0. (c) x3-3x2 -4x+1=0, 751. Divide a hemisphere of radius 1 into two equal parts by a plane parallel to the base. 752. Evaluate the positive root of the equation x' 5x 3 = 0 to within 0.0001. 753. Compute to within 0.0001 the root of the equation: (a) x4 +3x3-9x 9=0 lying in the interval (1, 2); (b) x4 -4x3+4x2 -4=0 lying in the interval (-1, 0); (c) x4 +3x3+4x2 +x-3=0 lying in the interval (0, 1); (d) x4 -10x2 -16x+5=0 lying in the interval (0, 1); (e) x4x' 9x2 + 10x 10 =0 lying in the interval (-4, 3); (f) x4 -6x2 +12x-8=0 lying in the interval (1, 2); (g) x4 -3x2 +4x-3=0 lying in the interval (-3, 2); (h) x4 x3 -7x2 -8x-6=0 lying in the interval (3, 4); (i) x4 3x3 + 3x2 2 =0 lying in the interval (1, 2). 754. Compute to within 0.0001 the real roots of the following equations: (a) x4 + 3x3-4x 1 =0, (b) x4 +3x3 x2 -3x+1=0, (c) x4 -6x3+13x2 -10x+1=0, (d) 4 8x3-2x2 +16x-3=0, (e) x4 -5x3 +9x2 -5x-1=0, (f) x4 -2x3-6x2 +4x+4=0, (g) x4 +2x3+3x2 +2x-2=0, (h) x4+4x3-4x2-16x-8=0.
-
CHAPTER 6 SYMMETRIC FUNCTIONS Sec. 1. Expressing Symmetric Functions in Terms of Elementary Symmetric Functions. Computing Symmetric Functions of the Roots of an Algebraic Equation 755. Express the following in terms of the elementary symmetric polynomials: (a) x?+ 4-3xixax,, (b) x?x2 + xix3+.4x3+ xixR + x3x, + x2x3, (C) x4 + x2 + 4-244-244-244, (d)44+ x1 x2 + 4x3 + xi x3 + 4x3 + (e) (xi +x2) (xi +x3) (x2+x,), (f) (x?+.4) (4+ xN) (4+4), (g) (2x1 X2 X3) (2x 2 X3) (2X3 X2), (h) (x1 x2)2 (x 1 x3)2(X2 X3)2. 756. Represent the following in terms of the elementary symmetric polynomials: (a) (xi +x2) (xi +x3) (x, +x4) (X2 + X3) (X2 + X4) (X3 + X4),
(b) (XIX 2 + X3X4) (X1X3 + X 2X 4) (XIX 4 + X2X3),
(c) (xi+X 2 X3 X4) (X1 X2 + X3 X4) (X1 X2 X3 + X4).

757. Represent the following monogenic polynomials in terms of the elementary symmetric polynomials: (a) 4+ , (g) x?x2x3+ , (n) 4x2x3x4+ . . . , 3X2 2 (h) x;x2x3+ , (o) Xi (b) 4+ , X3 + , 3 3 (i) 44+ , (C) XjX2X3 + Xi X2 +, (p) (d) .4.4+ -, (J) 4x2 + , (q) -Xi 4X2X3 + , (e) .4x2+ , (k) 4+ ., (r) 44+ ,
4 (f) Xi + , (1) (m)
XX3X3X4 + . . . , 4.4X3 + ,
(s) 4x2 X2 + , (t)
4+
CH. 6. SYMMETRIC FUNCTIONS
117
758. Express the following in terms of the elementary symmetric polynomials: (a) ( +x2+ x3 + +x,,)2 +(xix2 +x3 + . . . +x)2+ (xi +x2x3 + +x)2 + + (x, +x2+ xa + x)2, (b) (xi +x,+x2 + ...+x) (x1x2 +x, + 759. Express the following in terms of the elementary symmetric polynomials:
(a) (c)
E (xi -x02,
i>k
(b)
i>k
+ X03,
(xi xkr,
i>k
(d)
E
1>k lOi;1#k
(xi +xk -x. ,)2.
760. Express the following monogenic polynomial in terms of the elementary symmetric polynomials:
xDc3
, + xl 2
761. Express the following in terms of the elementary symmetric polynomials:
E (a1xh -f-a2xi2 -1- +au x-4)2.

The sum is extended over all possible permutations i1, i2, of the numbers), 2, ..., n. 762. Express the following in terms of the elementary symmetric polynomials: (a) x i X2 X3 X2 X3 .4_ Xl
X2 x3 Xi Xi X2X3 + (X2 X3)2 (X3 X2+ X3 + \
X1)2
(b)
(c)
(X1 X2)2
Xi+ X2 / f X2 +
X3+ + X2 + X2 ).
1 X3 Xi / 1 xi X2 X3 / \ X2 763. Express the following in terms of the elementary symmetric polynomials: (a) xixi_ +x1x3 x2x3 X2X4 X3X4 x3x4 x2x4 X1X2 X2X3 XIX," x1X3
(b) Xl+X2 Xi+X3 + Xi+X4 + X2+X3 + X2+X.2 + X3+X4
X3 + X4 X2 + X4 X2 + X3Xl+ Xri x 1+ X3
X1 + X2.
118
PART I. PROBLEMS
764. Express the following in terms of the elementary symmetric polynomials: (a) (d)
E. 4 ,= , E f_q_, Jo., x. .,
(b) (e)
Ex
( '
c)
E :a, x. io;
,
E XP 'of ,
(f) E xk xi i0 .1
i#k j>k
765. Compute the sum of the squares of the roots of the equation x3+2x 3 = 0. 766. Compute xix2 +x1x3+x3x3+x2x3+.4x1+x3x? of the roots of the equation x' x2 4x + 1 =0. 767. Determine the value of the monogenic symmetric function
4x2x3+
of the roots of the equation x4+ x3 2x2 3x + 1 = 0. 768. Let xi, x2, x3be the roots of the equation x3+px+q =O. Compute: (a) +++++ X2 X3 Xj. x2 x1
Xi X2 X3 X2 X3 Xi
,
x3
(b) xf4+ 44 + 44 + 4.4 + 4.4+ x34, (c) (4 x2x3)(4 xlx,)(xi x2x1),

(d) (xi+ x2)4(x1'+3)4 (X2 +X3)4,
(e)
xi
(X2 + 1) (X2+ 1)
x2
(X1+ 1) (X3+ 1)
xi
(Xi+ 1) (X2 + 1)
4 (x1+1 )2 (X2 + 1)2 (X3 + 1)2 ' 769. What relationship is there between the coefficients of the cubic equation x 3+ax2+bx+c=0 if the square of one of the roots is equal to the sum of the squares of the other two? L' 770. Prove the following theorem: for all roots of the cubic equation x3-Fax2+bx+c=0 to have negative real parts, it is necessary and sufficient that the following conditions hold: a > 0, abc>0, c>0. (f
119
771. Find the area and the radius of a circle circumscribed about a triangle whose sides are equal to the roots of the cubic equation ,3 -ax2 + bxc= 0. *772. Find the relationship among the coefficients of an equation whose roots are equal to the sines of the angles of a triangle. 773. Compute the value of a symmetric function of the roots of the equation f (x) = 0 : (a) xlx2+ . . . , f (x) = 3 x-.3 5x2 + 1; (b) + . . . , f (x)=3x4 2x3+ 2x2+ x 1;
+ 4)(4+ x2x3 + (x3+ x,xi+ f (x)= 5x3 6x2 +7 x 8. 774. Express in terms of the coefficients of the equation a ,f x3+ aix2+ a 2x + a3=-0
the following symmetric functions:
(c)
(a) 4 (x1 x9)2 (x1 x3)2 (x2 x3)2,
(b) 4(4- x 2 x3)(4- x1 x3)(4- x1x2),

(c)
(x1- x2)2 (xi- x2)2 (x2- x.)2
x2xs (d) a3(4+ x1x2 + 4)(4 + x2x3 + x3)(x3+ x,x, + x?).

xsxs 775. Let x1, x2, ..., x be the roots of the polynomial xn + + + an. Prove that the symmetric polynomial in x 2, x3, . x can be represented in the form of a polynomial in x1. 776. Using the result of Problem 775, solve Problems 755(e), 755 (g), 774 (b), 774 (d).
777. Find
=1 function of x1, x2, ..., 778. Let the expression of the symmetric function F (x1, x 2, ..., x,) in terms of the elementary symmetric functions be known. Find the expression of
al ;k where fk is the kth elementary symmetric uxi
E
i =1
a!in terms of the elementary sym-
metric functions.
120
PART I. PROBLEMS
Prove the theorems: 779. If F(xi, x 2, ..., xn) is a symmetric function having the property
F (xi+ a, x 2+ a, ..., xn + a)= F
x 2,
x,)
and if 0:13. (f1, f 2,..., f) is its expression in terms of the elementary symmetric functions, then
do
dfi
+ (n 1)f, P,5-- +
uJ 2
fn-i
ao n ofn
and conversely. 780. Every homogeneous symmetric polynomial of degree two having the property of Problem 779 is equal to a E (xi xk)2
i<k
where a is a constant. 781. Find the general form of homogeneous symmetric polynomials of degree three having the property of Problem 779. 782. Using the result of Problem 779, express the following in terms of the elementary symmetric polynomials:
i<pck
783. Prove that among the symmetric polynomials F(xi, x 2, xn) having the property
(x, - xj)2 (x1 xk)2 (xj xk)2.
F (x1, x2, ..., x)= F (x,+ a, x2+a,
xn + a)
there are n-1 "elementary polynomials" y2, y3, ..., y such that each polynomial of the class under consideration can be expressed in the form of a polynomial in y 2, c03, (P. 784. Express the following symmetric functions in terms of the polynomials y2, y, of Problems 783: (a) (x1 x,)2 (x1 x,)2(x2 x3)2, (b) (x1 x2)4 + (Xi X3)4+ (X2 X3) 4. 785. Express the following symmetric functions in terms of the polynomials y2, y3, p4of Problem 783: (a) (x1+x2 x3 x 4) (x1 x2+ xa x4) (x1 x2 x3+x4), (b) (x1 x2)2 (x1 x3)2 (x1 x4)2(x2 x3)2(x2 x4)2(x3x4)2.
121
Sec. 2. Power Sums 786. Find an expression for s2, Sd, S4, S5, s6in terms of the elementary symmetric polynomials, using Newton's formulas. 787. Express f2, f 4, f5, f6 in terms of the power sums sl s , using Newton's formulas. 788. Find the sum of the fifth powers of the roots of the equation x6 4x5+ 3x3 4x2 +x + 1 =0. 789. Find the sum of the eighth powers of the roots of the equation .X4 1 =0. 790. Find the Sum of the tenth powers of the roots of the equation x3 -3x+1 =0.
,
791. Find s1, s 2,

Xn +
s of the roots of the equation xn- 2 1 + 1 1 2 + + n =O.
792. Prove that
k [bk k bk -2ac + ak (xi` + 4)=( l)
k(1 2 bk-4a2c2 k (k 4)(k 5) bk-6 a3c3

1. 2.3
3)
if xl, x2are the roots of the quadratic equation ax2+bx+c=-0. 793. Prove that for any cubic equation
S3
3 (fif2)
794. Prove that if the sum of the roots of a quartic equation is equal to zero, then
S5 Ss S2
5
57
3
S5
2
S2
795. Prove that if s5 =s3 =0 for a sextic equation, then

7 5 2
796. Find nth-degree equations for which s,=s2 ==4_1 =0. 797. Find nth-degree equations for which
Sa = 53= ...=Sn=0.
122
PART 1. PROBLEMS
798. Find an nth-degree equation for which s2= 1, s3=s4= =sn=s.+1=0. 799. Express
Exk 4 in terms of power sums.

<3 <j
*800. Express E (x1 -1- xi)k in terms of power sums. *801. Express E 1<j xf )2k in terms of power sums.
f, 1
f1 f2 fk-i fk-2 1 0
0
1
2f2 802. Prove that sk = 3f3

kfk
Sl
fi

1 803. Prove that fk = k i
S2sl S3 s2
2
Si. 3

0
0
sk
Sk-1
Sk 2
Si 1 0
xn
sl
xn 1
1
sl
xn
0
804. Compute the determinant
S2
Sn
Sn_i Sn -2
*805. Find sm of the roots of the equation
X (x)=0.
*806. Prove that f 2, f3and f 4of the roots of the equation X (x) = =0 can only take on values 0 and 1. *807. Solve the system of equations
F x2 + -F 4+4+ -1-x,z 2 =a,
x? +x2+ +x,, n=a and find x7+1 + 4+1 + +4+1
123
*808. Compute the power sums sl, s2, the equation

.X" + (a + b)
sn of the roots of
xn-1+(a2 +ab+b2) .Xn-2 -.0. + +(an+an-lb+ +bn)s of the roots of
*809. Compute the power sums sl, s2, the equation x n + + b) xn-i + (a2 + b2) xn-2
+ (an bn)=_. O.
Sec. 3. Transformation of Equations

810. Find equations whose roots are:
(a) xi + x2, x2 + x3, x3-Fx1; (b) (x1 x2)2, (x2X3)2, (X3 X1)2 ;
(C) X?
x2x3,
x3x1,
xix2;
(d) (x1 x2 ) (x1 x3), (x2 x1) (x2 x3), (x3 x1) (x3 x 2); 4, 4 (e) x?, 4, 4; f ( ) where xi, x2, x3are the roots of the equation x3 +ax2+bx+ c =O. 811. Find an equation whose roots are (xi +x2 e+x, e2)3 and (xi +x2 e2 +x3 0 3 "1/ I where e 2 + i 2 ; x1, x2, x3are the roots of the equation
x3+ax2+bx+c =O.
812. Find an equation of lowest degree, one of the roots of
which is 11 - + 2+x2, where xi, x2, x3are the roots of the cux2 x2 x1 bic equation x3+ ax2+bx+ c= 0, and the coefficients of which are expressed rationally in terms of the coefficients of the given equation. 813. Find an equation of lowest degree, one of the roots of xi which is where x1, x2, x3are the roots of the equation x3+ X2 +ax2+bx+c=0, and the coefficients of which are expressed in terms of the coefficients of the given equation. 814. Find an equation of lowest degree with coefficients expressed rationally in terms of the coefficients of a given equation
124
PART I. PROI3Ltis AS
x4+ ax3+ bx2+ cx+ d= 0, one of the roots of the desired equation being:
Sa) x1x2 + x3x4, (b) (xi+ x2 x3 x4)2, (c) x1x2, [(d) xi +x2, (e) (x1 x2)2. 815. Using the results of Problems 814 (a) and 814 (b), express the roots of a quartic equation in terms of roots of the auxiliary cubic equation of Problem 814 (a). [816. Write a formula for the solution of equation x4 6ax2 + bx 3a2 =0. 1817. Write an equation, one of the roots of which is (x1x2+x2x3+x3x4+x4x5+x5x1) x (x1x3 +x3x5 +x5x2 +x2x4 +x4x1) where x1, x2, x3, x4, x5are roots of the equation x5+ax+b=0.
Sec. 4. Resultant and Discriminant

14'818. Prove that the resultant of the polynomials (x) ...+an and qo (x)= box'n + ...+b, is equal to a determinant made up of the coefficients of the remainders left after dividing co (x), xcp (x), xn -1 cp (x) by f (x). It is assumed that the remainders are arranged in order of increasing powers of x (Hermite's method). !Remark. The remainder rk (x) left after dividing xk -1cp (x) by f (x) is equal to the remainder obtained upon division of xrk _i(x) by f (x). *819. Prove that the resultant of the polynomials f (x)=a0xn + aixn-1+ and cp (x)= boxn + bixn-1+ .+b is equal to a determinant composed of the coefficients of polynomials of degree n-1 (or lower)
Yk
ixk-2 + + ak-i)p(x) (x) (aoxk + a (boxk-1+bi xk-2+ n (Bezout's method).
+bk _i)f(x)
k = 1,
125
Remark. 41=aoy bo f,
`Pk = X4k-1+ ak-1p
bk-1J
*820. Prove that the resultant of the polynomials
f (x)= aoxn + aixn-1+ + a. and

cp (x)= box'n + bixm '+ + bn,
is equal, for n> m, to a determinant made up of the coefficients of polynomials xk(x) of degree not exceeding n 1 determined from the formulas
Xk (X)= xk-l p (X)
for 1
+ ak,+.-1)
m,
m (x) (bock -n +in-1
Xk (x)=
(aoxk-n+m-i
avck-n+m-2
+ bixk-n-Fm-2+
bk-n+m-1)f(X)
(the polynomials Xkare arranged in order of increasing powers of x). Remark. Xn-.+1=a0xn-m cp (x) bo f 00,
Xk = XX1,-1+ak-n+m--1 Xn-m (x) bk_n+m_if(X) for k>nm+ 1.

821. Compute the resultant of the polynomials:
(a) (b) (c) (d) (e) (f)
xs 3x2 + 2x + 1 and 2x2x 1 ; 2x3-3x2 +2x+ 1 and x2 +x+3; 2x33x2 x + 2 and x4 -2x2 -3x+4; 3x3+2x2 +x+ 1 and 2x3 +x2 x-1; and 3x3 xs + 4 ; 2x4x3+ 3 and box2+bix+b2. a,x2+a1x + a2
822. For what value of X do the following polynomials have a common root: and xs + Xx + 2; (a) x3 Xx+2 (b) x3 -2Xx+ X3and xs + X2 -2; (c) xs+ Xxs 9 and x3 +Xx-3? 823. Eliminate x from the following systems of equations:
(a) xs xy +ys = 3,
xsy +xys = 6 ;
126
PART I. PROBLEMS
(b) x3 xy y 3+ y=0, (c) y =x3 2x2 6x+8,
x2 +xy2 -1=0; y=2x3-8x2 +5x+2.
824. Solve the following systems: (a) y2 7xy + 4x2+13x-2y-3=0, y2 -14xy+9x2 +28x-4y 5=0; (b) y2+ x2 y 3x = 0, y2 6xy x2 + lly + 7x 12 = ; (c) 5y2 -6xy+5x2 -16=0, y2 xy + 2x2 y x 4 = 0 ; (d) y2 +(x-4)y+x2 -2x+3=0, y3-5y2 +(x +7)y+x3x2 -5x-3=0; (e) 2y3 -4xy2 (2x2 12x + 8)y +x3+ 6x2 16x =0, 4y3 (3x+10)y2 (4x2 -24x+16)y-3x3 + 2x2 12x +40 =0. 825. Determine the resultant of the polynomials anxn +aixn -i+ +an and ao yn aixn - 2 + _1. 826. Prove that 9i (f, cpl92) = %(.f, Pi) 9i. (f, P2). *827. Find the resultant of the polynomials Xn and xrn-1. *828. Find the resultant of the polynomials X, and X. 829. Compute the discriminant of the polynomial: (a) x3 x2 -2x+1, (c) 3x3+3x2 +5x+2, (e) 2x4 x3-4x2 +x +1. (b) x3 +2x2 +4x+1, (d) x4 x3 3x2+ x + 1,
830. Compute the discriminant of the polynomial: (a) x5-5ax3+5a2xb, (b) (x2 x + 1)3 X (x2 x)2, (c) ax3 bx2+ (b 3a) x + a, (d) x4 Xx3+ 3 (X-4) x2 2(X 8) x-4.
CH, 6. SYMMETRIC FUNCTIONS
12?
831. For what value of X does the polynomial have multiple roots: (a) x3-3x+ X; (b) x4 -4x+ A, (c) x3-8x2 + (13 A) x(6 +2A), (d) x4 4x3 + (2 A) x' +2x 2? 832. Characterize the number of real roots of a polynomial with real coefficients by the sign of the discriminant: (a) for a third-degree polynomial; (b) for a fourth-degree polynomial; (c) in the general case. 833. Compute the discriminant of the polynomial xn +a. *834. Compute the discriminant of the polynomial xn+px+q. *835. Compute the discriminant of the polynomial aoxin+ n a ixm + a2. 836. Knowing the discriminant of the polynomial + an aoxn + aix n 1 + find the discriminant of the polynomial
an xn + a_ ixn-i +
+ a,.
837. Prove that the discriminant of a fourth-degree polynomial is equal to the discriminant of its Ferrari resolvent (Problem 814(a) and Problem 80). 838. Prove that
D ((x a) f (x))= D (f(x)) [ f(a)]2

*839. Compute the discriminant of the polynomial + 1. xn-1 + x n-2 + *840. Compute the discriminant of the polynomial
xn+axn-1+axn-2+
+ a.
841. Prove that the discriminant of a product of two polynomials is equal to the product of the discriminants multiplied by the square of their resultant. 842. Find the discriminant of the polynomial
X
P
XP
zen-1 .-1 1
*843. Find the discriminant of the cyclotomic polynomial X.

128
PART I. PROBLEMS
*844. Compute the discriminant of the polynomial x x2 En =n!(1+ 1- + .2+ + 4). : 1 *845. Compute the discriminant of the polynomial F.= x.a x_ + a (a 1)(a n+ 1) 2 1 ) xn a(a2+
n1
*846. Compute the discriminant of the Hermite polynomial x2 xR

P (x) = ( )ne
2
2 due dxn
*847. Compute the discriminant of the Laguerre polynomial P (x)= ( 1)n dn (dne . *848. Compute the discriminant of the Chebyshev polynomial 2 cos (n arc cos I ). *849. Compute the discriminant of the polynomial
1).
dn (
( x)
x2r +1
1 -1- x2 )
dxn
*850. Compute the discriminant of the polynomial P (x)= ( 1)n (1

n+ X2) 2
dn
VI
1
X2
dXn
*851. Compute the discriminant of the polynomial
p.(x)= (
x2n+2e--;
dn+ (e x )
dxn+
*852. Find the maximum of the discriminant of the polynomial xn + aixn + + a all the roots of which are real and connected by the relation x; + +4=n (n 1) R2. 853. Knowing the discriminant of f (x), find the discriminant of f (x9. 854. Knowing the discriminant of f (x), find the discriminant of f (xn").
129
855. Prove that the discriminant of F (x)= f (cp (x)) is equal to
[D (f)1"
fl D(cp (x) xi),
where m is the degree of cp (x); x1, x2, ..., x are roots off (x). The leading coefficients of f and cp are taken equal to unity. Sec. 5. The Tschirnhausen Transformation and Rationalization of the Denominator 856. Transform the equation (x -1) (x 3) (x + 4)=0 by the substitution y =x2 x 1. 857. Transform the following equations:
(a) x3 -3x -4=0 by the substitution y = x2+ x + 1 ; (b) x3+2x2+2=0 by the substitution y =x2 + 1; (c) x 4 X 2=0 by the substitution y =x3 2; (d) x4 x3 x2 + 1 = 0 by the substitution y = x3 +x2 +x + 1 .
858. Transform the following equations by the Tschirnhausen transformation and find the inverse transformations: y=x2+x; (a) x3 x+2 =0,
y=x3+x; (c) x 4+5x3+6x2 1 =0, y =x3+ 4x2 +3x 1.

(b) x43x + 1 =0,
-
859. Transform the equation x' x2 2x +1=0 by the substitution y=2 x2 and interpret the result. 860. Prove that for the roots of a cubic equation with rational coefficients to be expressed rationally with rational coefficients in terms of one another, it is necessary and sufficient that the discriminant be the square of a rational number. 861. Rationalize the denominators: (a)
1
1-FV-2-1/
(b)
(c)
1-42-1-21 14
1V2-1-
862. Rationalize the denominators: (a) +1' th, 2- 3a -1 c 0-1-2oc+1 '

oc3 -3oc+ 1
= 0;
+ oc2 3cx + 4 = 0;
5. 1215
130 1 (c) 3e-1-0-2m1

(d) oc3 -1-3a2 -1-
PART I. PROBLEMS
0(4
cc3 +2oc + 1 = 0; 4a2 3a 2 0.
+2'
oc4
863. Prove that every rational function of a root x1of the Axi + B cubic equation x3 + ax2+bx+ c=0 can be represented as
Cxi + D
with coefficients A, B, C, D, which can be expressed rationally in terms of the coefficients of the original expression and in terms of the coefficients a, b, c. 864. Let the discriminant of a cubic equation that has rational coefficients and is irreducible over the field of rationals be the square of a rational number. It is then possible to establish the +13 among the roots. What condition do the relation x2 = yx,-1- coefficients a, (3, y, a have to satisfy? 865. Make the transformation y =x2 in the equation aoxn +aixn-l+ +an = O. 866. Make the transformation y =x3in the equation aoxn+aixn-'+ +an = O. *867. Prove that if all the roots xi of the polynomial f (x)-- xn+aixn-i+ +a,, an 0 with integral coefficients satisfy the condition I xi S 1, then they are all roots of unity. Sec. 6. Polynomials that Remain Unchanged under Even Permutations of the Variables. Polynomials that Remain Unchanged under Circular Permutations of the Variables 868. Prove that if a polynomial remains unchanged under even permutations and changes sign under odd permutations, then it is divisible by the Vandermonde determinant made up of the variables, and the quotient is a symmetric polynomial. 869. Prove that every polynomial that remains unchanged under even permutations of the variables can be represented as F1+F, A where F1and F2are symmetric polynomials and A is the Vandermonde determinant made up of the variables.
131
870. Evaluate 1 1 1
2 XI XI
4-2
n +I
X2 a2 x
X 2+1 +I X2
xn+I n
n 2 X2 n xn
871. Form an equation whose roots are axi + (3x2 +yx3, ocx2+ + [3x3 yx,and ocx3 +3x1+yx2 where x1, x2, x3 are the roots of the equation x3+axe + bx + c = O. 872. Form an equation whose roots are x,y,+x2y2 +x3y2, x1y2 +x2y3 +x,),1, x1y3 +x2y1+x3y2 where x1, x2, x3are the roots of the equation x3+px+q +0, and yl, y2, y3are the roots of the equation y2+ p' y + g' =O. 873. For the following equations with rational coefficients x3 +px+ q =0, y3+ p'y q' =0 to be connected by a rational Tschirnhausen transformation, it is necessary and sufficient that the ratio of their discriminants A and A' be the square of a rational number and that one of the equations
u3 = 3pp'u +
27 qq VAA
'
have a rational root. Prove this. 874. Prove that every polynomial in n variables x1, x2, x which remains unchanged under circular permutations of the variables may be represented as 11, Air11 2 1.02where 71,,
...,
7 12 71,,
are linear forms:

= xlz
+ x2e2 +
+ xn,
= x10
x2s4 +
27
+ Xny
xn
xie n 1 + x26.2n 2
.
E= COS 7
stn
The exponents al, oc2, oc_, satisfy the condition: n divides cf.1 +2a2 + + (n I) 875. For rational functions that do not change under circular permutations of the variables, indicate n elementary ones (frac.
132
PART I. PROBLEMS
tional and with nonrational coefficients) in terms of which. all of them can be expressed rationally. 876. For rational functions of three variables unaltered under circular permutations, indicate three elementary functions with rational coefficients. 877. For rational functions of four variables that remain unchanged under circular permutations, indicate four elementary functions with rational coefficients. 878. For rational functions of five variables that remain fixed under circular permutations, indicate five elementary functions with rational coefficients.
CHAPTER 7 LINEAR ALGEBRA
In this chapter we adhere to the following terminology and notations. The term space is used to denote a vector space over the field of real numbers, unless otherwise stated. This term is used both for the space as a whole and for any part of a larger space (the term subspace will be used only when it is necessary to specify that a given space is part of a larger space). A linear manifold is a set of vectors of the form X0 + X, where X0is some fixed vector and X runs through the set of all vectors of some subspace. The equation X --(x1, x2, ..., xn) means that X has coordinates x1, x2, xi, in some fixed basis of the space; when we deal with Euclidean space, the basis is assumed to be orthogonally normalized. Vectors are sometimes called points, one-dimensional manifolds are called straight lines, and two-dimensional manifolds are called planes. Sec. 1. Subspaces and Linear Manifolds. Transformation of Coordinates 879. Given a vector space spanned by the vectors X1, X2, X,,,. Determine the basis and dimension:
(a) X1=(2, 1, 3, 1), X,=(- 1, 1, -3, 0);

(b) X1=(2, 0, 1, 3, -1),
X2 =(1, 2, 0, 1),
X2 =(1, I, 0, -1, 1),

-
13 = (0,
2, 1, 5, -3), X4 = (1, -3, 2, 9,
5);
(c) X1=(2, I, 3, - 1),

X2 = (4, 5, 3, - 1),
X2 = - 1 , 1,
3, 1),
X4= (1, 5, -3, 1).
134
PART I. PROBLEMS
880. Determine the basis and dimension of the union and intersection of spaces spanned by the vectors X1, Xk and Y1, Y.; (a) X1=(1, 2, 1, 0), Y1 = (2, -1, 0, 1), X2 = ( - 1, 1, 1, 1), Y2=(1, -1, 3, 7); Y1=(2, 5, -6, -5), (b) X1= (1, 2, -1, -2),
X2= (3, 1, 1, 1), Y2 = ( 1 , 2, -7, -3), X3=( -1, 0, 1, - 1); (c) X,=(1, 1, 0, 0), Y1= (0, 0, 1, 1), X2 =(1, 0, 1, 1), Y2 = (0, 1, 1, 0). 881. Find the coordinates of the vector X in the basis E1, E2, E3, E4: (a) X -(1,2, 1, 1), E1=(1, 1, 1, 1), E2 =(1, 1, -1, -1), E3 =(1, -1, 1, -1), E4 =(1, -1, -1, 1); (b) X= (0, 0, 0, 1), El= (1 , 1, 0, 1), E2 = (2, 1, 3, 1), / 1, -1, - 1). E3 (1, 1, 0, 0), E4 = (0, 882. Develop formulas for the transformation of coordinates from the basis E1, E2, E3, E4 to the basis Ei, 4 E73, (a) El=(1, 0, 0, 0), E2 = (0, 1, 0, 0), E3 = (0, 0, 1, 0), E4 = (0, 0, 0, 1), El= (1, 1, 0, 0), (1, 0, 1, 0), E'=(1, 0, 0, 1), E!,=(1, 1, 1, 1); (b) E1= (1, 2, -1, 0), E2 = (1, -1, 1, 1), / E3=( 1 - , 2, 1, 1), E4 = ( 1, - 1, 0, 1), E;= (2, 1, 0, 1), E=(0, 1, 2, 2), E'=( -2, 1, 1, 2), E4=(1, 3, 1, 2). 883. The equation of a "surface" with respect to some basis E1, E4 has the form x;+.4-.4-4= 1. Find the equation of this surface relative to the basis E;=(1, 1, 1, 1), E.=.(1, -1, 1, -1), E'2=(1, 1, -1, -1), E4= (1, -1, -1, 1) E4). (the coordinates are given in the same basis E1,
CH. 7. LINEAR ALGEBRA
135
*884. In the space of polynomials in cos x of degree not exceeding n, write the formulas for transformation of coordinates from the basis 1, cos x, cos x to the basis 1, cos x, ..., cos nx, and conversely. 885. Find a straight line in four-dimensional space that passes through the origin of coordinates and intersects the straight lines: x1=2+3t, x2 =1- t, x3= -1 +2t, x4 =3-2t and
x1=7t, x2=1, x3=1+ t, x4= -1+2t. Find the points of intersection of this straight line with the given straight lines. 886. Prove that any two straight lines in n-dimensional space can be embedded in a three-dimensional linear manifold. 887. Investigate, in general form, the condition for solvability of Problem 885 for two straight lines in n-dimensional space. 888. Prove that any two planes in n-dimensional space can be embedded in a five-dimensional linear manifold. 889. Give a description of all possible cases of the mutual location of two planes in n-dimensional space. 890. Prove that a linear manifold can be characterized as a set of vectors containing the linear combinations aXi + (1 -a) X2 of any two vectors X1, X2 for arbitrary a.
Sec. 2. Elementary Geometry of n Dimensional Euclidean Space
-
891. Determine the scalar product of the vectors X and Y: (a) X= (2, 1, -1, 2), Y=(3, -1, -2, 1); (b) X=(1, 2, 1, -1), Y=(-2, 3, -5, -1). 892. Determine the angle between the vectors X and Y: (a) X=(2, 1, 3, 2), Y-(1, 2, -2, 1); (b) X-(1, 2, 2, 3), Y= (3, 1, 5, 1); (c) X=(1, 1, I, 2), Y= (3, 1, -1, 0). 893. Determine the cosines of the angles between the straight line x1=x2 = =x and the axes of coordinates. 894. Determine the cosines of the interior angles of a triangle ABC which is specified by the coordinates of the vertices: A-(1, 2, 1, 2), B= (3, 1, -1, 0), C=(1, 1, 0, 1).
136
PART I. PROBLEMS
895. Find the lengths of the diagonals of an n-dimensional cube with side unity. 896. Find the number of diagonals of an n-dimensional cube which are orthogonal to a given diagonal. 897. In n-dimensional space, find n points with nonnegative coordinates such that the distances between the points and from the origin are unity. Place the first of these points on the first axis, the second, in the plane spanned by the first two axes, etc. Together with the coordinate origin, these points form the vertices of a regular simplex with unit edge. 898. Determine the coordinates of the centre and radius of a sphere circumscribed about the simplex of Problem 897. 899. Normalize the vector (3, 1, 2, 1). 900. Find the normalized vector orthogonal to the vectors (1, 1, 1, 1); (1, 1, 1, 1); (2, 1, 1, 3). 901. Construct an orthonormal basis of a space, taking for two vectors of this basis the vectors
1
, (
1
2 ,
1
2 - )
2 - ,
and ( 6
1 ,
16)
902. By means of the orthogonalization process, find the orthogonal basis of a space generated by the vectors (1, 2, 1, 3); (4, 1, 1, 1); (3, 1, 1, 0). 903. Adjoin to the matrix (1 1 2 1 0 1 1 0 1 2 1 0
1 2 2
two mutually orthogonal rows that are orthogonal to the first three rows. 904. Interpret the system of homogeneous linear equations an
a21 + an X2
+ 2 x2 + . . . +Rln x=0, . . . a2 x,,-0, +a, x--0
a mi xi +a m2 x2 +
and its fundamental system of solutions in a space of n dimensions, taking the coefficients of each equation for the coordinates of a vector.
137
905. Find an orthogonal and normalized fundamental system of solutions for the system of equations 3x1- x2 - x3+ x4= 0, x1 + 2x2 - x3 x 4 = 0. 906. Decompose the vector X into a sum of two vectors, one of which lies in a space spanned by the vectors A1, A2, . and the other is orthogonal to this space (the orthogonal projection and the orthogonal component of the vector X): (a) X=(5, 2, -2, 2), Ai =(2, 1, 1, -1),
A2 " A ' (1 5 1, 3, 0); (b) X=(-3, 5, 9, 3), Ai =(1, 1, I, 1),
A2=(2, -1, 1, 1), A3= (2, -7, -1, -1). A1, A2, . A. to be linearly independent, give formulas for computing the lengths of the components of the vector in Problem 906 when posed in general form. 908. Prove that of all vectors of a given space P, the smallest angle with a given vector X is formed by the orthogonal projection of the vector X on the space P. 909. Find the smallest angle between the vectors of the space P (spanned by the vectors A1, A,) and the vector X:
907. Assuming the vectors
(a) X= (1, 3, -1, 3), A,=(I, -1, 1, 1), A,= (5, 1, -3, 3); (b) X=(2, 2, -1, 1), Ai =(1, -1, 1, 1), A,=(- 1, 2, 3, 1), A3=(1, 0, 5, 3). 910. Find the smallest angle formed by the vector (1, 1, ..., 1) with the vectors of some rn-dimensional coordinate space. 911. Prove that of all vectors X- Y, where X is a given vector and Y runs through a given space P, the vector X- X', where X' is the orthogonal projection of X on P, is of smallest length. (This smallest length is called the distance from the point X to the space P.) 912. Determine the distance from the point X to the linear manifold A 0 + ti Ai + + t,A (a) X=(1, 2, -1, 1), A0 =(0, -1, 1, 1), A1=(0, -3, -1, 5), A2 = (4, -1, -3, 3); (b) X=(0, 0, 0, 0), A,=(1, 1, 1, 1), Ai =(1, 2, 3, 4).
138
PART I. PROBLEMS
913. Consider a space of polynomials of degree not exceeding n. The scalar product of polynomials f1, f2is defined as
f f1(x) f2(x) dx. Find the distance from the origin to the linear
manifold consisting of the polynomials xn +a, xn + ...+ an. 914. Indicate a method for determining the shortest distance between the points of the two linear manifolds X,+ P and Yo + Q. 915. The vertices of a regular n-dimensional simplex (see Problem 897), the length of an edge of which is unity, are partitioned into two sets of m + 1 and n-m vertices. Linear manifolds of smallest dimension are passed through these sets of vertices. Determine the shortest distance between the points of these manifolds and determine the points for which it is realized. *916. Given, in a four-dimensional space, two planes spanned by the vectors A1, A2 and B1, B2. Find the smallest of the angles formed by the vectors of the first plane with the vectors of the second plane :
(a)
(1, 0, 0, 0), B2=(2, -2, 5, 2);
= (0, 1, 0, 0), B1=(1,1, 1, 1), (1 , I, 1, 1),
(b) A, --- (1 , 0, 0, 0), A2=(0, 1, 0, 0), B2 =(1, -1, 1, -1).
*917. A four-dimensional cube is cut by a three-dimensional "plane" passing through the centre of the cube and orthogonal to a diagonal. Determine the shape of the solid obtained in the intersection. *918. Given a system of linearly independent vectors B1, B2, B .. The set of points made up of the endpoints of the vectors t,B, + t2B2 + + t ,B 05 t1 1, ..., 0 t n, 1, is called a parallelepiped constructed on the vectors B1, B2, . B .. Determine the volume of the parallelepiped inductively as the volume of the "base" [B1, B2, ..., .13,,,_ ]] multiplied by the "altitude" equal to the distance from the endpoint of vector B, to the space spanned by the base. The "volume" of the one-dimensional "parallelepiped" [B1] is considered equal to the length of the vector B1. (a) Develop a formula for computing the square of the volume and assure yourself that the volume does not depend on the numbering of the vertices. (b) Prove that V [cBi, Bz, B ,]=[ c V [B1, Bz, BR,].
139
(c) Prove that V [1 .3+B' ', B2, B] V [13, B2, ..., B m ]+ B .] and determine when the equal sign holds + V [ifi', B2, true. ;919. Prove that the volume of an n-dimensional parallelepiped in n-dimensional space is equal to the absolute value of the determinant made up of the coordinates of the generating vectors. *920. Let C1, C2, ..., C mbe the orthogonal projections of the vectors B1, B2, ..., B. on some space. Prove that
V [Ci, C2, ..., C m]

*921. Prove that
V [Bi, B2,
V [A1, A2, ..., A., B1, ..., Bk] 5 V [A1, ..., A .] V [B1, ..., Bk] (cf. Problem 518). 922. Prove that V [Ai,
A2, ...,
A m ]I
A11 . 1A21 1
A mi
(cf. Problem 519). 923. Find the volume of an n-dimensional sphere using Cavalieri's principle. 924. Consider the space of polynomials whose degree does not exceed n. For the scalar product we take f f1 (x) f2(x) dx. Find the volume of the parallelepiped formed by the vectors of the basis relative to which the coefficients of the polynomial are its coordinates. Sec. 3. Eigenvalues and Eigenvectors of a Matrix 925. Find the eigenvalues and eigenvectors of the following matrices: 2 1 0 a\ (a) ( 1 2 ) ' (b) 4 2 ' (c) 7 a 0 j ' /1 / \1 1 1 1 1 1 I 1\ 1 1 ' 1' 5 (e) 1 1 6 0 2
(d)
3 1 ), 1
140 (2 (f) 5 1 (3 (i) 4 4 1 3 0 1 1 8
PART I
PROBLEMS
2 3 , 2 0 0 , 2
001 (g) 0 1 0 , 1 00 ( j) 2 4 3 5 6 6
(h)
0 2 2 0 1 3
3 , 0
6 9 8
926. Knowing the eigenvalues of the matrix A, find the eigenvalues of the matrix A-1. 927. Knowing the eigenvalues of the matrix A, find the eigenvalues of the matrix A2. 928. Knowing the eigenvalues of the matrix A, find the eigenvalues of the matrix Am. 929. Knowing the characteristic polynomial F (X) of the matrix A (of order n), find the determinant of the matrix f (A), where
f (x) = b0(x 0 (x U . .. (x ,,,)

930. Knowing the eigenvalues of the matrix A, find the determinant of the matrix f (A), where f (x) is a polynomial. 931. Knowing the eigenvalues of the matrix A, find the eigenvalues of the matrix f (A). 932. Prove that all the eigenvectors of the matrix A are eigenvectors of the matrix f (A). *933. Find the eigenvalues of the matrix 1 / 1
1
1 s
2
E(n-1)
1
E2
e4
e2 (n-1)
. . . e2 (n-1)
-1)'
\1
where c = cos - i sin 7` n an odd number. n n *934. Find the sum 1 + ,

935. Find the eigenvalues of the matrices:
x y (a) y 0 y
x\ x 0 ... x ,
0
x ..
7 a,
(b)
an
a2 a1
an an-1)
\y y y
\a, a3
141
1 0 1 1 0
1
(c)
0 1 1 0 /
*936. Knowing the eigenvalues of the matrices A and B, find the eigenvalues of their Kronecker product. 937. Prove that the characteristic polynomials of the matrices AB and BA coincide for arbitrary square matrices A and B. 938. Prove that the characteristic polynomials of the matrices AB and BA differ solely in the factor ( X)n-m. Here, A is a rectangular matrix with m rows and n columns, and B is an n-by-m matrix, n > m. Sec. 4. Quadratic Forms and Symmetric Matrices 939. Transform the following quadratic forms to a sum of squares: (a) 4 +2x1x, + 24 + 4x, x, + 54, (b) 4x1 x,+ 2x1x, +'44 + x3,
(c) x, x2 + x2 x3+ x3 xi, (d) x? - 2x, x, + 2x, x, 2x, x4 + 4 + 2x, x, 4x, x, + 4 24, (e) x?+ x, x2 + x, x4. 940. Transform the quadratic form
Xi Xk i<k
to diagonal form. 941. Transform the quadratic form xi x, I

i<k
to diagonal form. 942. Prove that all the principal minors of the positive quadratic form are positive.
142
PART I. PROBLEMS
*943. Let the quadratic form
= an + a 12 x1 x2 + . . . + alnX1x + a21 X2 X1 a22 + + a2 x2 x

+
x x1-I- ant Xn X2
...
a x;;
+an..x.,c2
be reducible to the diagonal form cci xi2 +a2 the "triangular" transformation
= bi2 X2 + . . .
x;2 +
by
bin xn
b2n An
X2=
x.
It is required to: (a) express the coefficients al, a2, a in terms of the cm-fficients a,k ; (b) express the discriminants of the forms fk (x1+1, x) = = x12 42 in terms of the coefficients ad,. Find the condition under which a triangular transformation of the indicated type is possible. 944. Prove that the necessary and sufficient condition for positivity of the quadratic form
f = anx?+ a12_xi x2+ . . . + ain a1x a21X2 Xi + a22 x2 + . . . + a x2 xn + a1x,, x1+ antx x2+ . . . + an
is fulfilment of the inequalities
an a12 a11 a 12
an > 0;
a22
> 0; . . ;
a21
a1,, a22 a2,, >0
a22
ant a2 (Sylvester's condition). *945. Prove that if to a positive quadratic form we add the square of a linear form, the discriminant of the former increases. *946. Let f x2, ..., xn)= au x?+ ... be a positive quadratic
form,
p (x2, ..., x)= f (0, x2,
x)
143
Df
and D, their discriminants. Prove that
Df ( all
947. Let
f(x,, x 2, ..., x)=1T+ n+ . . . + 42-1,241-1,2+2
l2+0
where 11,12, ..., 1,, 4+1,4+2, ...,1,+, are real linear forms in x1, x2, . . x. Prove that the number of positive squares in a canonical representation of the form f does not exceed p, and the number of negative squares does not exceed q. *948. Let so, s1, ... be power sums of the roots of the equation xn + a, xn-1+ ...+ a= 0 with real coefficients. Prove that the number of negative squares in a canonical representation of the quadratic form E
k =1
Si+k+2 Xi Xk
is equal to the number of pairs
of conjugate complex roots of the given equation . Prove the following theorems : 949. Fulfillment of the following inequalities is a necessary and sufficient condition for all the roots of an equation with real coefficients to be real and distinct:
So S1 S2
So
so S 1
0, S2 S2 S3 54 51 S2 S3
>0; ...; A=
S1
S2 . . . So
> 0.
Sn-1 Sn Stn-2
*950. If the quadratic forms

f= all x j + a12 x1 X2
. . .
a, a22 x2 . + a2n x2 x + a2
Xn X2 + + a,; A n
+ a21 x2
+ an1 xn and =b x? +b, 2 x,.X2

. . . bln x1Xn
b21 X2 Xi + b22
+ b2n X2 An
+ bn1x xl+ b2x xo + . . . + b A
144
PART I. PROBLEMS
are nonnegative, then the form
(f, so) = all b11 xi + a12 bi Xi X2 + + a21 b21 X2 X1 + a22 b22 x2 +
4- aln bin X1 Xn
+ azn b2n x2 x.
i + an1b, x x, + an, b2x x2+ . . . + abxl

is nonnegative. 951. Transform the following quadratic forms to canonical form by an orthogonal transformation: (a) (b) (c) (d) 2x? + x2 4x1 x, 4x2 x 3, x?+ 2x3 + 3x3 4x, x, 4x, x,, 3, 3x? + 4x2 + 5x3 + 4x1 x2 4x2 x
2x? + 5.4+ 54+ 4x1x, 4x1 x, 8x2 x,, (e) xi 2x2 2x3 4x1x, + 4x1 x,+ 8x2 x2, (f) 5x, + 6x2 + 4x3 4x1x, 4x1x,, (g) 3x? + 6x2 + axi 4x1x, 8x1 x, 4x2 x3, (h) 7x? + 5x2 + axi 8x1X2 8X2 X3, 4x, x, + 2x, x4 + 2x, x, 4x, x4, (i) 2x? + + 2x3 +
(j) 2x1X2
+ 2.X:3 x4,
2x1 x2 - 2x1 xi - 2x2 x, 2x3 x4, 2x2 x4 + 2x3X4, (1) 2x1x2 2x1 x3 2x1 x4 2X2 (m) x? + x2 + x3 + x?, - 2x1x, + 6x1x, 4x1 x4 4x2 x3+ 6x, x4 2x3 x 4, (n) 8x1X3 + 2X1 X4 + 2X2 X3 + 8x2 X4.
(k) x? + x2 + x3 + 952. Transform the following quadratic forms to canonical form by an orthogonal transformation: (a) E x?+
i= I i<k
xi x
(b) E xi xk
i<k
953. Transform the form X1x,+x, x3 + ...+xn _ l x to canonical form by an orthogonal transformation. 954. Prove that if all the eigenvalues of a real symmetric matrix A lie in the interval [a, b], then the quadratic form with mat-
145
rix A A E is negative for A> b and positive for A < a. The converse holds true as well. 955. Prove that if all the eigenvalues of a real symmetric matrix A lie in the interval [a, c] and all the eigenvalues of a real symmetric matrix B lie in the interval [b, d] then all the eigenvalues of the matrix A + B lie in the interval [a+ b , c+ d]. 956. Let us call the positive square root of the largest eigenvalues of the matrix A A (A is a real square matrix, A is its transpose) the norm of the matrix A and denote it by 11 A II. Prove that
(a) I A 11=11 A II, (b) I AX I I AII' I X I;

(c) A+B A II+ B
the equality holds for some vector X0,
(d) II ABII<IIA II IIB 11, (e) the moduli of all eigenvalues of the matrix A do not exceed I A II. 957. Prove that any real nonsingular matrix can be represented as a product of an orthogonal matrix and a triangular matrix of the form
b11 b12 b22
. bin I bt,'
with positive diagonal elements b1 and that this representation is unique. 958. Prove that any real nonsingular matrix is representable in the form of a product of an orthogonal matrix and a symmetric matrix corresponding to some positive quadratic form. 959. Let there be a quadric surface in n-dimensional space given by the equation 2 + . . . + a, xi x all + an xi x + x2 xi + a22X2 + . . . + a, A 2 x + a2 xn x2 + . . . + ant, 4,
+ a1x
+2b1 x1+2b 2 x2 + . . . + 2b + c = 0
146
PART 1. PROBLEMS
or, in abbreviated notation, AX.X+2BX + c=0. Prove that for the centre of the surface to exist it is necessary and sufficient that the rank of the matrix A be equal to the rank of the matrix (A, B). 960. Prove that the equation of a central quadric surface may be reduced to canonical form
y=0 a, x; + . . . + ocr +
by a translati on of the origin and by an orthogonal transformation. 961. Prove that the equation of a noncentral quadric surface may be reduced to canonical form
al + ar 2 kr r + 1
by a translation of the origin and by an orthogonal transformation. Sec. 5. Linear Transformations. Jordan Canonical Form 962. Establish that the dimension of a subspace into which the entire space is mapped under a linear transformation is equal to the rank of the matrix of this linear transformation. 963. Let Q be a subspace of dimension q of the space R of dimension n, and let Q' be the image of Q under a linear transformation of rank r of the space R. Prove that the dimension q' of space Q' satisfies the inequalities min (q, r). 964. Using the result of Problem 963, establish that the rank p of the product of two matrices of ranks r1 and r2 satisfies the inequalities r,+r2 n,..p. min (r,., *965. Let P and Q be any complementary subspaces of the space R. Then any vector X E R decomposes uniquely into a sum of the vectors Y E P and Z E Q. The transformation consisting in going from vector X to its component Y is called projection on P parallel to Q. Prove that projection is a linear transformation and its matrix A (in any basis) satisfies the condition A2 = A. Conversely, any linear transformation whose matrix satisfies the condition A2 = A is a projection.
Cli. 7. LINEAR ALGEBRA
147
*966. The projection is termed orthogonal if P l Q. Prove that in any orthonormal basis, the matrix of orthogonal projection is symmetric. Conversely, any symmetric idempotent matrix of the same degree is a matrix of orthogonal projection. *967. Prove that all nonzero eigenvalues of a skew-symmetric matrix are pure imaginaries, and the real and imaginary parts of the corresponding eigenvectors are equal in length and orthogonal. *968. Prove that for a skew-symmetric matrix A it is possible to find an orthogonal matrix P such that
0 a1 a, 0 0 a, -- a, 0
P -1,4P=
0 a, a, 0 0
0 (all elements not indicated are zero; a1, a2, ..., a, are real numbers). 969. Prove the theorem : if A is a skew-symmetric matrix, then the matrix (E A) (E + A)-' is an orthogonal matrix without 1 as eigenvalue. Conversely, every orthogonal matrix that does not have 1 as an eigenvalue can be represented in this form. *970. Prove that the moduli of all eigenvalues of an orthogonal matrix are equal to 1. *971. Prove that eigenvectors of an orthogonal matrix which belong to a complex eigenvalue are of the form X+ i Y, where X, Y are real vectors equal in length and orthogonal. *972. Prove that every orthogonal matrix can be represented as
Q-1TQ,
148
PART 1. PROBLEMS
where Q is an orthogonal matrix and T is of the form / cos cp, sin cp, cos cp, sin co, cos cp2 sin cp2
Sill (p2
cos cp2
1 1
(all other elements being equal to zero). 973. Reduce the following matrices to the Jordan normal form : 1 (a) ( 0 2 13 (c) ( 5 6 4
(e)
2 2 2 16 7 8 2 3 1 8 10 8 10
0 0 1
4 , (b) 3 3 3 , (d) 3 2 7 (f) ( 3 2 0 , (h)

(
6 5 6 0 1 0 12 4 0 3 8 14
0 0 1 8 6 5 2 0 2 , , ,
16 6 7
4 3
7) , 7 6 6 4
2 (g) ( 4 4
1 2
3 6 ), 10
J-L 7. LINEAR ALGEBRA
149
1 21 26 5 2 1 22 4 16 1 3 2
1 17 21 2 1
1
8 , (j) 6 6 3 2 4 1 3 2
30 19 23 7 5 10
(i) ( 5 6 (4 (k) 2 1 (9 (m) 1 8 1

(o) (
14 9 11 3 2 , 3
(1)
6 1 5 1 3 2
(n)
1 3 2
2 6 4
3 2
974. Reduce the following matrices to the Jordan normal form: 3 (a) 1 1 1 6
. .
4 7 \ 17
0 0 2 1
0 0 , 1 0'
1 0 (b) 0 \0
2 1 0 0
3 2 1 0
4 3 2 1 ,1
/0
0 (c)
1
0
0 . o 1 ... 0 0 0
. . .
0 \1
0 0
. .
1 Ot
*975. Prove that any periodic matrix A (satisfying the condition An1=E for some natural m) is reducible to the diagonal canonical form. *976. Knowing the eigenvalues of the matrix A, find the eigenvalues of the matrix A,c, composed of appropriately arranged mth-order minors of the matrix A (see Problem 531). 977. Prove that any matrix A can be transformed into its transpose. *978. Prove that any matrix can be represented as a produc t of two symmetric matrices, one of which is nonsingular.
150
PART L PROBLEMS
979. Starting with a given matrix A of order n, construct a sequence of matrices via the following process : A1 = A, tr A1=Pi, A, pl E=B1 , B,A=A2,
R2
2 tr A 2 =p2, A2 p 2 E=B2,
-
A=
---
Ao=iy A p trA =p, ApE=B
B_,A=A,
n i
where tr Ai is the trace of matrix A. (the sum of the diagonal n 2, , pn are the coefficients of the chaelements). Prove that .1,, racteristic polynomial of the matrix A written in the form (_ l)n [Xn p1 Xn _p2Xn 2 p]; matrix B is a zero matrix; finally, if A is nonsingular, then -- B_,= Pn *980. For the equation X Y YX= C to be solvable in terms of square matrices X, Y, it is necessary and sufficient that the trace of the matrix C be zero. Prove this.
PART II. HINTS TO SOLUTIONS
11. See Problem 10. 13. Demonstrate the validity of the theorem for each of the four operations on the two numbers and take advantage of the method of mathematical induction. 18. Use the fact that the left members are easily represented as a sum of two squares. 27. Set x=a+bi, y=c+di. 28. Set z=cosp+i sin cp. 31. Set z= t2, z'= t'2. Use Problem 27. 37. Go over to the trigonometric form. 38. 1+ (,) = 40. Pass to the half-angle. 1 41. Convince yourself that z= cos 0 + i sin 0 ; z= cos 0 T i sin 0. Take advantage of De Moivre's formula. 51. Set a= cos x+ i sin x. Then cos' " x= 2 )2m , etc.
52. Show that the coefficient of (2 cos x)' -2p is equal to ( 1)' ,Cfn _ p+ +CP-1 )'Take advantage of the method of mathematical induction. 53. This is similar to Problem 52. 54. Make use of the binomial expansion of (I +i)^. 55. Use Problem 54. 1/-3 56. Expand ( 1 +i ---: I nusing Newton's binomial formula. 3 68. Show that the problem reduces to computing the limit of the sum 1+1 1 +a-Fa2 + ..., where a= 2 1 cos 2a 69. Take advantage of the fact that sine a 2 2 71. Use the fact that cos 3a 3 cos a 3 sin a sin 3a coss a= ,sin 3 a= + 4 4 4 4
152
I- and 1+ 72. In computing sums of the type 1 + 2a+ 3a2 + + na"+22a+33a3 +...+ Haan-1it is useful first to multiply them by 1 - a. x3 =a0+ Po.), ce8+ (3.= -q, 34= -p. 76. xi =a+P, x2 =aw+ 77. Multiply by -27 and regard the left member as the discriminant of some cubic equation. 78. Set x=a+P. 87. Show that e"= -1. 277 2n , then the desired sum can be written as 1+ 88. If e = cos - + sin n n +,2+ ...+,n -1. 89. Consider two cases: (1) k is divisible by n; (2) k is not divisible by n, 91. 92. Multiply by 1- e. 94. (a) Subtract from the sum of all 15th roots of I the sum of the roots belonging to the exponents 1, 3, and 5. 97. The length of a side of a regular 14-sided polygon of radius unity is 4-tr 7-c equal to 2 sin - Use the fact that cos - + i sin - satisfies the equation x6 + 7 7 1 4' +x5 +x4 +x3 +x2 +x+1 =O. 98. (1) If x1, x2, ..., xnare roots of the equation aoxn+aix"-i+ ...+an =0, then aoxn + aixn + ...+ an= ao (x- xi) ...(x- xn). (2) If e is an nth root of 1, then 6, the conjugate of e, is also an nth root of 1. 99. In the identities obtained from Problem 98 set x=1. 100. Take advantage of the factorization of x"- 1 into linear factors. 101. In the factorization of x" - 1 into linear factors set: (1) x= cos 0+ + i sin 0, (2) x= cos 0- i sin 0. 103. Take advantage of the fact that the moduli of conjugate complex numbers are equal. 105. (a) Reduce the equation to the form 107. Let S= cos cp + Cn i cos (y+ x)x+... +cos (cp T -= sin cp + Cn i sin (cp + tux) xn , x +.. + sin (cp + me) xn.
x+
x-
1r
=1
Compute S+ Ti and S- Ti and determine S from the resulting equations. 113. First prove that cp (p)=13' (1- 1), if p is a prime number. To do this, count the numbers not exceeding I," that are divisible by p. -1 and only such roots are not primi116. Prove that all roots of xPm 1 tive roots of xPm - 1. 117. Show that if n is odd, then to obtain all the primitive roots of degree 2n of unity it is sufficient to multiply all primitive nth roots by -1. 119. Use Problem 118. 120. Use Problems 115, 116, 1 1 1 and show that (1) (p)= -1 if p is prime; (2) that p. (p")= 0 if p is prime, a> I, (3) v. (ab)= (a) p. (b) if a and b are relative prime. 2lor 2k7c belongs to the exponent n1, 122. Show that if ek =cos - + i sin n then x- ek will enter the right member of the equation being proved to the 1n . power El,. (d1), where d1runs through all divisors .7
CI-I. 2. EVALUATION OF DETERMINANTS
153
123. Consider the cases: (I) n is the power of a prim. (2) n is the product of powers of distinct primes. For Case (1) use Problem 11 for (2) use Problems 119 and 122. 124. Consider the cases: (I) n is odd and exceeds 1; (2) 21 ; (3) n=2n1, n1is odd and exceeds 1; (4) n=21cni, where k> 1, n1is odd and exceeds 1. 125. Use the identity
Xi X2 +X1 X3+ +Xn_ Xn
(x,+x, +
+x), (xl +x2 +
+ x;;)
2
Consider the cases: (1) n is odd; (2) n=2121, n1is odd; (3) n =21' ni, where k> 1, Ili is odd. 126. Multiply the sum S by its conjugate and take into account that ex' does not change when x+ n is substituted for x.

132. Bear in mind that each pair of elements of a permutation constitutes an inversion. 133. The number of inversions in the second permutation is equal to the number of orders in the first. 145. Show that each term has 0 for a factor. 149, 150. Replace rows by columns. 153. Find out how the determinant will change if its columns are permuted in some fashion. 154. (a) Note that when x= ai, the determinant has two identical rows. 155. To the last column add the first multiplied by 100 and the second multiplied by 10. 156. First subtract the first column from each column. 163. Subtract the first column from the second. 179. Add the first row to all other rows. 180.182. Subtract the first row from all other rows. 183. Subtract the second row from all other rows. 184. Add the first row to the second. 185. Add all columns to the first. 186, 187. From the first column subtract the second, add the third, etc. 188. Expand by elements of the first column or add to the last row the first multiplied by xn, the second multiplied by xn-1, etc. 189. Add to the last column the first multiplied by xn-the second multiplied by xn -2, etc.
154
PART 11. HINTS TO SOLUTIONS
190. Construct a determinant equal to f (x+ 1)-f (x). In the resulting determinant subtract from the last column the first, the second multiplied by x, the third multiplied by x2, etc. 191. Multiply the last column by a1, a2, ..., anand subtract, respectively, from the first, second, ..., nth column. 192. Add all columns to the first. 194. Add all columns to the last one. 195. Take a1out of the first column, a2out of the second, etc. Add to the last column all the preceding columns. 196. Take h out of the first column; add the first column to the second, 197. Multiply the first row and the first column by x. 198. Subtract from each row the first multiplied successively by a1, a21 N a,,. From each column subtract the first multiplied successively by a1, a2, 199. Add all columns to the first. 200. Add to the first column all the others. 201. From each column, beginning with the last, subtract the preceding column multiplied by a. 202. From each row, beginning with the last, subtract the preceding row. Then to each column add the first. 203. Multiply the first row by b0, the second by b1, etc. To the first row add all succeeding rows. 204. Take a out of the first row and subtract the first row from the second. 205. Expand by elements of the first row. 206. Represent as a sum of two determinants. 208. Add a zero to each off-diagonal element and represent the determinant as a sum of 2" determinants. Use Problem 206 or 207. 211. Multiply the first column by xn - 1, the second by xn-2, etc. 212. Expand by elements of the last column and show that An =xAn-i+ + anxix .x,,_ , (andenotes a determinant of crder a). Use mathematical induction in computing the determinant. 213. Expand by elements of the last column and show that An+1= xnAn+ 214. Take a1out of the second column, a2out of the third, ..., and an out of the (n+ 1)th. Reverse the sign of the first column and add all columns to the first. 215. Expand in terms of elements of the first row. 216. Expand in terms of elements of the first row and show that An- i = 219. Use the result of Problem 217. 221. Expand by elements of the first row and show that A n = xAn _ 1-An- 2. 222. From the last row subtract the second last multiplied by--y"- Yn - 1
+ anY
.Y,,.
Show that An =
Yn
Y n -1
(Xn Y n
Xn -1 Yn) An-i
223. Represent as a sum of two determinants and show that An= anAn-i+ aia2 an_ 225. Represent as a sum of two determinants and show that An= (an -4 An_1+x (a1-x) ... (a _ 1- x). 226. Setting x= (x,,- a)+ an, represent the determinant in the form of a sum of two determinants and show that
An = (xnan) At1-1+ an (x1-a1) (x 2 -a2) (xn-1an-1).
155
227. Represent in the form of a sum of two determinants and show that
An= (xn -anbn)
An_1+ an bn (x1- b1) . . (xn_1-an 1bn 1) 228. Represent in the form of a sum of two determinants and show that
-
An=
An_i+(_
230. Expand by elements of the first row and show that
Aar = (a2-b2) Au, -2 231. From each row subtract the preceding one and add to the second all subsequent rows. Then, expanding the determinant by elements of the last row, show that An=[a+ (n-1) b] An _ 2+ a (a+ b) ... [a+ (n- 2) b]. 232. Represent as a sum of two determinants and show that n-1 An = x (x - 2a ) An _1+ a;2, xn-1 (x - 2a,). 2, represent the determinant as a sum 233. Setting (x- a)2 = x (x-2a)+an of two determinants and show that An = x (x -2a) A n _ i + (x -2a1 ) .. (x - 2an -1) 234. Represent as a sum of two determinants and show that An =An _ 1+(b.,. ...
235. Represent the last element of the last row as an-an. Pr ove that
An =(- 1)n-1b1b2... bn_ l an-
An _ i.
236. From each row subtract the next. 237. Set 1 =x1 (1 x) in the upper left corner. Represent the determi-
nant as a sum of two determinants. Use the result of Problem 236. 238. Multiply the second row by x1-1, the third by xn -2, ..., the nth by x. From the first column take out xn, from the second, xn ...,from the nth, x. 239. Use the suggestion of the preceding problem. 240. From each column subtract the preceding one (begin with the last column). Then from each row subtract the preceding one. Prove that An = k --An _ 1. When calculating, bear in mind that Cnk= ck _1 + Cn 241. From each column subtract the preceding one. 242. From each row subtract the preceding one. Prove that An =An _,. 243. Take m out of the first row, m+1 out of the second, ..., m+n out 1 1 of the last. Take - out of the first column, out of the second, k k+ 1 etc. Repeat this operation until all elements of the first column become equal to 1. 244. From each column subtract the preceding one. In the resulting determinant, subtract from each column the preceding one, keeping the first two fixed. Again, subtract from each column the preceding one, keeping the first three columns fixed, and so on. After m such operations we get a determinant in which all elements of the last column are 1. The evaluation of this determinant presents no special difficulties.
156
i = 245. From each row subtract the preceding one and show that = (x 1) An. = 246. From each row subtract the preceding one and show that = (n 1)! (x-1) An. 247. From each row subtract the preceding one; from each column subtract the preceding one. Prove that An = aA,. 248. Represent the last element of the last row as z+(xz). Represent the determinant as a sum of two determinants. Use the fact that the determinant is symmetric in y and z. 249. See the suggestion of Problem 248.
a ab out of the first column and subtract from suiting determinant, take a (cc p)
252. Subtract from each row the first row multiplied by -. In the re-
the first column all the other columns. 253. Add all columns to the first and from each row subtract the preceding one. See Problem 199. 254. Use the suggestion of Problem 253. 256. Regard the determinant as a polynomial in a of degree four. Show that the desired polynomial is divisible by the following linear polynomials in a:
a+b+c+d, a+bcd, ab+cd, abc+d.

258. Adding all columns to the first, separate out the factor x+ al + ... +an Then setting x=a1, a2, ..., an, convince yourself that the determinant is dix an. visible by x xa2, 259. The Vandermonde determinant. 264. Expand by elements of the first column. 265. From the second row subtract the first. In the resulting determinant subtract the second row from the third, etc. 1 269. Take out of the third row, out of the fourth, etc. i i 2 270. Make use of the result of Problem 269. 271. Take 2 out of the second column, 3 out of the third, etc. When computing Ii
n?i>
k3
(i2 k2) it is useful to represent (i2 k2) =11 (i k) II (i + k) .
X1 X2 out of the first column out of the second, etc. x1 1 ' x 2 1 273. Take a7 out of the first row, a; out of the second, and so on. 275. To the first column add the second multiplied by On, the third multiplied by Cl, etc. 276. Take advantage of the result of Problem 51. 277. Take advantage of Problem 53. 278. Adjoin the row 1. x,, x2, ..., x and the column 1, 0, 0, ..., 0. 272. Take
CI-I. 2. EVALUATION OF DETERMINANTS
157
279. Consider the determinant I x,

D = .7c1
1 xn
z
2.2
x2 . . xn xn n
x7 x1 2 1
zn
Compare the expansion of D by elements of the last column with the expression D=
H ( x; _xk)
i> k -1
1=1
n ( z_.,).
280. Use the suggestion of Problem 279. 282. Adjoin the first row I, 0, ..., 0 and thefi rst column 1, 1, 1, ..., I. Subtract the first column from all the succeeding columns. 285. Expand by elements of the last row. 286. First, from each column (beginning with the last) subtract the preceding column multiplied by x. Then, after reducing the order and taking out obvious factors, transform the first rows (dependent on x) using the relation (m+ os_ms=sms-i+ s(s2 1
) m
s-2 + ,.. 1.
287. From each column, beginning with the last, subtract the preceding column multiplied by x. 288. (m) Add to the first column the sixth and the eleventh, to the second column, the seventh and the twelfth, ..., to the fifth column, the tenth and the fifteenth. Add to the sixth column the eleventh, to the seventh column, the twelfth, ..., to the tenth column, the fifteenth. From the fifteenth row subtract the tenth, from the fourteenth row subtract the ninth, ..., from the sixth row subtract the first. 293. Consider 1 1 1 qao Cna1 C
. . . ag
bn
b? b7 -1
..
.
bn tz
C a . . . Wiz
b8 -1
o n an
. . . ai n 't
294. Consider sin a, cos a, 0 ... 0 sin a, cos a, 0 ... 0 cos a, cos 2 . cos en sin , sin a, . .. sin an 0 0 0
. . .
sin an cos an 0 .
0
.
1)
158 295. Consider

1 X1 I X2
. . .
PART 11. HINTS TO SOLUTIONS
1
x
1 1
xn
-1 x, . . . x2 . . . x2-1
0 0
xn
. . . xi nz
xn
1 0
0 . .. 0
296. Raise to the second power. 297. Subtract from the third column the first, from the fourth the second. Then multiply by cos y sin cp 0 0 sin cp 0 0 0 0 sin 2y cos 2<p
cos cp 0 cos 2cp 0 sin 2cp
298. Subtract from the second column n times the first, from the fourth, n times the second. Interchange the second and third columns. Multiply by cos ncp sin lip sin ny cos ncp 0 0 0 0 0 0 cos (n+1) cp sin (n+1) cp 0 0 --sin (n+1) cp cos (n+1)(p
299. Square it. Transform as a Vandermonde determinant and transform each difference to the sine of some angle. This will yield the sign. 300. Study the product
as
an -1
a, ao
a2 . an -1 a1 . . . an-2
1 1
1
1 r.
... 1

en -1
01
az
a3 ... no
e7 -1 ... et,z,"
where e k = COS
2kn
12
+i
sin
2krc
308. Consider e, = cos 7r + i sin - - . Then
n ( 2
kO
ak 0)
1 1
CH. 4. MATRICES
159
311. Use Problem 92. 314.

k=0 n- I
(ao +
k+ a2 q+ + (72n - 1 ckn --I )
= 11 Rao + 0,0+ (al.+

r=0 n-1
ao +0 a + + (an _i+ 02n _ 1) 07-1]
X 1-1 Rao a) + (ai an+ 1) Ps + . . + (an _1 02n _

s= 0
2rr; . . 2r 7c kit , . kit where ck=e0S ---H-t sin ; o,, = cos +1 sin ; 0 n 0 n (2s+ 1) n (2s+ I)rc 3s cos +isin
n n
323. From each row subtract the first, from each column subtract the first. 325. Use Problem 217. 327. Represent in the form of a sum of determinants or set x=0 in the determinant and its derivatives. 328. (1) From the (2n-1) th row subtract the (2n 2)th, from the (2n-2)th row subtract the (2n-3)th, ..., and from the (n+ 1)th row subtract the nth, from the nth row subtract the sum of all the preceding ones. (2) Add to the (n+ i )th row the ith, i=1, 2, ..., n-1. 329. Add to every row all the subsequent ones, and subtract from every column the preceding column. Prove that An +1(X) = (X- n) (x-1).
CHAPTER 4 MATRICES
466. Use the result of Problem 465 (e). 473. Consider the sum of the diagonal elements. 491. Take advantage of the results of Problems 489, 490. 492. Use the result of Problem 490. 494, 495. Use the results of Problems 492, 493. 496. Argue by induction with respect to the number of columns of the matrix B, first having proved that if the adjoining of one column does not change the rank of B, then it does not change the rank of the matrix (A, B) either. A proof other than by induction can be carried out by using the Laplace theorem. 497. Take advantage of the results of Problems 496, 492. 498. From the matrix (EA, E+ A) select a nonsingular square matrix P and consider the product (E-- A) P and (E+ A) P. 500. Take advantage of the result of Problem 489. 501. Prove the uniqueness of the representation in Problem 500 and thus reduce the problem to counting the number of triangular matrices R with a
160
given determinant k. Denoting the desired number by F,,(k), prove that if k= a b for relatively prime a, b, then Fn(k)= F,, (a) F,, (b). Finally, construct, inductively, a formula for Fn UP% where p is prime. 505. Take advantage of the results of Problems 495, 498. Find the matrix P with the smallest possible determinant so that P AP is diagonal, and then use the result of Problem 500. 517. Use the Laplace theorem and the Bunyakovsky inequality. 518. Establish the equation I AA = I BB I I CC I on the assumption that the sum of the products of the elements of any column of matrix B into corresponding elements of any column of matrix C is equal to zero. Then complete (in appropriate fashion) the matrix (B, C) to a square matrix and take advantage of the result of Problem 517. 523. On the left of the determinant, adjoin a column, all elements of which are equal to 2 adjoin at the top a row, all elements of which (except the corner) are 0; then subtract the first column from all other columns. 527. Take advantage of the results of Problems 522, 526. 528. Establish a connection between an adjoint matrix and an inverse matrix. 529. With respect to the minor formed from elements of the first m rows and the first m columns of the adjoint matrix, establish the result by considering the product of the matrices / An Am+1,1 An1 Al2 Am + 1,2 11112 / au a12 . am
\
Aim
..
Am-FL m 1
Arun
a21
a22
a2n
ant
ant
afirl
I / where Aik are the cofactors of the elements aik. Do the same for the general case. 535. Represent A x B as (A x E,,,) (E,,x B). 537. First analyze the case when A11is a nonsingular matrix and then argue by induction. Reduce the general case to this case, adding AE to the matrix.
CHAPTER 5 POLYNOMIALS AND RATIONAL FUNCTIONS OF ONE VARIABLE

547. (a) Expand f (x) in powers of x-3, then substitute x + 3 for x. 553. Differentiate directly and substitute x=1, then isolate the maximum
power of x and continue the differentiation. 555. Consider the polynomials (x)= of (x) xf ' (x), f2 (x)= nfi(x) xf (x) and so on.
CH. 5. POLYNOMIALS ANT) FUNCTIONS OF ONE VARIABLE
161
561. Prove by the method of mathematical induction. 562. The nonzero root of multiplicity k 1 of the polynomial f (x) is a root of multiplicity k 2 of the polynomial xf (x), a root of multiplicity k 3 of the polynomial x [xf ' (x)]' , etc. Conversely, the general nonzero root of the polynomials f (x), xf ' (x), x [xf' (x)]' , . (a total of k 1 polynomials) is a root of f (x) of multiplicity not lower than k 1. 563. Differentiate the equation showing that the polynomial is divisible by its derivative. 567. Consider the function fi (x) or 12 (x) fi (x) f2 (x) 568. Relate the problem to a consideration of the roots p (x)=f (x)f '(x0)f' (x) f (x2) where xo is a root of [f (x)]2 f (x) f" (x). 569. Use the solution of Problem 568 and expand f (x) in powers of x x0. 576. Prove it like d'Alembert's lemma. 580, 581. Represent the function in the same form as in proving the d'Alembert lemma. fk a ) f (z)= f (a) + k`l (a)= 0 . (za)k [1+ co (z)], 583. Find the roots of the polynomials and take into account the leading coefficients [in Problems (a) and (b)]. It is advisable, in Problem (c), to set x= tan20 when seeking the roots. 589. Find the common roots. 608. First prove that f (x) does not have any real roots of odd multiplicity. 623. Use the result of Problem 622. 626. Use the fact that the equation should not change when x is substi1 tuted for x and for x. - is substituted for x and 627. The equation should not change when 1 x 1x for x. 637. Divide by (1 x)n and differentiate m-1 times, assuming x=0 after each differentiation. Take advantage of the fact that the degree of N(x) is less than m and the degree of M (x) is less than n. 642. Use the Lagrange formula. Perform the division in each term of the result and collect like terms using the result of Problem 100. 644. Express f (x0) in terms of f (xi), f (x,), f (xd, using the Lagrange interpolation formula and compare the result with the hypothesis of the f (xd. problem, taking into account the independence of f (x,), f (x,), Then study cp (x)=(xx1) (xx2)...(xxd, expanding it in powers of x x0. 645. Represent the polynomial xg in terms of its values by means of the Lagrange interpolation formula. 648, 649. Construct an interpolation polynomial by Newton's method. 650. Find the values of the desired polynomial for x=0, 1, 2, 3, ... , 2n. 651. The problem can be solved by using Newton's method. A shorter way is to consider the polynomial F (x) = xf (x) 1, where f (x) is the desired polynomial. 652. Consider the polynomial (x a) f (x) I.
6. 1215
162
653. Construct the polynomial by Newton's method and, for conveni-
en ce of computation, introduce a factorial into the denominator of each term. 654. Consider the polynomial f (x2), where f (x) is the desired polynomial.
655. The easiest way is by the Lagrange formula
f (x)
cp (x)
(x1,
x2, ,
v
k= I
f (xk)
(x - xk) cp' (Xk)
are roots of the denominator). 656. First expand by the Lagrange formula, then combine complex conjugate terms.
Xn
657. (e) Use Problem 631. (f) Set
a+x
= y. (d), (h) Seek expansions
by the method of undetermined coefficients. Find part by the substitution x =x1, x2, ..., xnafter multiplying by the common denominator. Then differentiate and again set x = xi, x2, xn 1 into 660. Use Problem 659. In Problem (b) decompose x2 - 3x +2 partial fractions. 665, 666. Take advantage of Problem 663. 667. In Problem (c) expand the polynomial in powers of x-1. 668. Expand in powers of x-1 (or put x=y +1). 669. Set x =y+ 1 and use mathematical induction to prove that all coefficients of the dividend and divisor (except the leading coefficients) are divisible by p. 670, 671. The proof is like that of the Eisenstein theorem. 679, 680. Assuming reducibility of f (x), set x= al, a2, ..., an and draw a conclusion concerning the values of the divisors. 681. Count the number of equal values of the presumed divisors. 682. Use the fact that f (x) does not have real roots. 683. Prove that a polynomial having more than three integral roots cannot have for one of its values a prime in the case of an integral value of the independent variable; apply this to the polynomial f (x) -1. 684, 685. Use the result of Problem 683. 702. Construct a Sturm sequence and consider separately the cases of even and odd n. 707-712. Derive recurrence relations between the polynomials of consecutive degrees and their derivatives and use them to construct a Sturm sequence. In Problem 708, construct a Sturm sequence solely for positive values of x and use other reasoning to assure yourself that there are no negative roots. In Problem 709, construct a Sturm sequence for negative x. 713. Use the fact that F' (x)=2f (x) f"' (x) and that f"' (x) is a constant. 717. Factor g (x) and use the result of Problem 716 several times. 718. Apply the result of Problem 717 to the polynomial xm. 719. Use the fact that if all the roots of the polynomial aoxn+aixn-2+ + an_ + anare real, then all the roots of the polynomial anxn+ an_ 3xn-i+ . +a0 are real. 721. Multiply by x-1. 727. Prove by contradiction by taking advantage of Rolle's theorem and the result of Problem 581.
163
728. Construct the graph of 9 (x)=
f (x)
f' (x)
and prove rigorously that
every root of [f' (x)]2 f (x) f " (x) yields an extreme point for 9 (x) and conversely. Prove that 4, (x) has no extreme points in the intervals, between the roots of f' (x), that contain a root of f (x), and has exactly one extreme point in the intervals which do not contain roots of f (x). 729. Use the result of Problems 727 and 726. 730. Study the behaviour of the function
(x) =
(x)+ x +X . f' (x) Y
731. It is solved on the basis of the preceding problem for A=0. 732. Prove by means of induction with respect to the degree of f (x), setting f (x) = (x+ A) fi (x), where L (x) is a polynomial of degree n-1. 733. The proof is obtained by a double application of the result of Problem 732. 734. If all the roots of f (x) are positive, then the proof is effected by elementary means, namely by induction with respect to the degree of f (x). Include in the induction hypotheses that the roots x1, x2, ..., x,,_1 of the polynomial b0+ blwx + ...+ b,,_ iw(n -1)2x"-' satisfy the condition 0 <x2 <x2 < ...<x,,_ 2 and xi>xi-2 w-2. To prove the theorem in the general case, it is necessary to represent wx' as the limit of a polynomial in x with roots not contained in the interval (0, n) and to take advantage of the result of Problem 731. 735. Consider
cp
(x)+4 (x)
cp
(x)-4 (x)
where
+ a" cos (9 + nO) x", (x) = a()cos cp + ti) (x)=14 sin cp + . . . +b" sin (9 + nO) xn
cp (x)
736. Consider the modulus of

cp
(x)+4 (x) where

-
(x) '
(x)= ao+a,. x+ +at, x", Li) (x)=b0+b1 x+ ...+b,,xn.
Having proved the real nature of the roots, multiply p (x)+i 4(x) by a pi and consider the real part. Use the result of Problem 727. 737. Decompose into partial fractions, investigate the signs of the cp fix) x coefficients in this decomposition and study the imaginary part i [cp (x) + 4 (x)] _ (x) . p(x) p (x) 738. Investigate the imaginary part of
f (x)
f (x)
by decomposing the
fraction into partial fractions. 739. Change the variable so that the given half-plane is converted into the half-plane Im (x) > 0. 740. Relate to Problem 739.
6*
164 741. Decompose
f' (x) into partial fractions and estimate the imaf (x)
ginary part. 743. Set x = yi and take advantage of the results of Problems 736 and 737. 744, 745. Take advantage of the result of Problem 743. y and use the result of Problem 744. 1 747. Multiply the polynomial by 1 x and, setting I xI=p> 1, estimate the modulus of (1 x) f (x). 746. Set x=
1 +y
CHAPTER 6 SYMMETRIC FUNCTIONS

772. The sides of a triangle similar to the given one and inscribed in a 1 circle of radius are equal to the sines of the angles of the given triangle. 2 800. First compute the sum
E ( x+x,)k i=i
and then substitute x xi and sum from 1 to n with respect to j. Finally, delete extraneous terms and divide by 2. 801. The solution is like that of Problem 800. 805. Every primitive nth root of unity raised to the mth power yields a primitive root of degree 11 -, where d is the greatest common divisor of m and n. As a result of this operation performed with respect to all primitive nth roots n of unity, all primitive roots of degree are obtained the same number of times. 806. Use the results of Problems 805, 117, and 119. 807. It is necessary to find an equation whose roots are x1, x 2, ..., xn. To do this, use Newton's formulas or a representation of the coefficients in terms of power sums in the form of a determinant (Problem 803). 808. The problem is readily solved by means of Newton's formulas or by means of representing power sums in terms of the elementary symmetric functions in the form of determinants (Problem 802). However, it is still easier to multiply the equation by (x a) (x b) and compute the power sums for the new equation. 809. The simplest way is to multiply the equation by (x a) (x b). 818. Consider the roots of the polynomial f (x) as independent variables. Multiply the determinant of the coefficients of the remainders by the Vandermonde determinant. 819. First prove that all polynomials 41, have degree n-1. Then multiply the determinant of the coefficients of 4k by the Vandermonde determinant. 820. The solution is like that of Problem 819.
165
827. Use the fact that the mth degrees of the primitive nth roots of 1 run through all primitive roots of unity of degree d, where d is the greatest common divisor of m and n. 828. Use the result of Problem 827 and the fact that R (X Xn) is a divisor of R xn 1) and R (Xn, xrn 1). 834, 835. Compute R(f' , f). 839. Multiply by x 1. 840. Multiply by x 1 and use the result of Problem 835. 843. Compute R (Xn, X,D. In computing the values of X;., for the roots of Xn, represent Xn in the form
) (xn 1) II (xa 1)
d considering that d runs through the proper divisors of n. 844. Take advantage of the relation E;,----En xn. 845. Take advantage of the relation 0. (nx x a) Fn x (x + 1) F,;+ - (a I) n l (a n) = 846. Use the relations
Pn= xPn_ i (n 1) P._ 29 P;1= nPni.
847. Use the relations
xn,= nPn+ n2Pn_ i, Pn= (x 2n + 1) Pn_ 1 (n-1)2 Pn_ 2.

848. Use the relations (4 x2) FL+ nxPn= 2n Pn_ 1, P,, xPn 1+ Pn_ 2=0. 849. Use the relations Pn-2x.P,+(x2 +1) P._ 2=0 , Pn= (n+ 1) Pn_i. 850. Use the relations
Pn (2n-1) xPn_ 1+ (n-1)2(x2+1) Pn 2=0, P;i= n2Pni.

851. Use the relations
Pn (2nx+
1) Pn_ i+ n (n-1) x2P_ 2= 0, p;i= (n+ 1) nPn_ 1

.
852. Solve the problem by Lagrange's method of multipliers. Write the result of equating the derivatives to zero in the form of a differential equation with respect to the polynomial that yields a maximum, and solve the equation by the method of undetermined coefficients. 867. First demonstrate that there is only a finite number of equations with the given properties for a given n. Then show that the properties are not destroyed under the transformation y=xm.
166

884. Use the results of Problems 51, 52. 916. The smallest angle is to be sought among the angles formed by vectors of the second plane with their orthogonal projections on the first plane. 917. Specify the cube in a system of coordinates with origin at the centre and with axes parallel to the edges. Then take four mutually orthogonal diagonals for the axes. 918. Use the result of Problem 907. 920. Prove by induction. 921. Use the fact that V [Ai, ..., Am, B1, BO= V [Ai, , A m ] V[B1, Bic] if A1 1B1 and use the result of the preceding problem. 933. First find the eigenvalues of the square of the matrix. Then, to determine the signs in taking the square root, use the fact that the sum of the eigenvalues is equal to the sum of the elements of the principal diagonal and that the product of the eigenvalues is equal to the determinant. Apply the results of Problems 126 and 299. 934. Apply the result of Problem 933. 936. Use the results of Problems 537 and 930. 943. (1) Use the fact that the determinant of a triangular transformation is equal to unity. (2) Set
...,
Xis +2=- Xk+ 2= =Xn=0.
945. For the new independent variable take the linear form whose square is added to the quadratic form. 946. Isolate one square from the form f and use the result of Problem 945. 948. Consider the quadratic form in the unknowns u1, u2, un:
n
f=E
cui + u2 xk+
k=1
where x1, x2, ..., x are roots of the given equation. 950. Decompose f and pinto a sum of squares and use the distributivity of the operation (f, pp). 965. In proving the converse, make use of the factorization X= AX+
+(E A) X.
966. Write the projection matrix in the basis obtained by combining the orthonormal bases P and Q. 967. Be sure that AX X=0 for any real vector X. Decompose the eigenvalue and eigenvector into a real part and an imaginary part. 968. Multiply the matrix A on the right by P, on the left by 13-1, where P
167
is an orthogonal matrix, the first two columns of which are composed of normalized real and imaginary parts of the eigenvector. 970, 971. Use the fact that for the orthogonal matrix A, AX AY= =X Y for any real vectors X and Y. 972. The proof is based on the results of Problems 970, 971 and, like Problem 968, on the results of Problem 967. 975, 976. Go to the Jordan canonical form. 978. Connect it with the solution of Problem 977. 980. For necessity, see Problem 473. For the sufficiency proof, consider first the case when all diagonal elements of the matrix C are zero. Then use the fact that if C= X Y YX, then S-1CS=(S-1XS) (S-1YS)(S-1 YS)(S-1XS).
PART III. ANSWERS AND SOLUTIONS
4 5 1. x= 11 ' Y.if 1 3 2. x= 2, y= z= 2, t= 2 . 3. I if n=4k; iif n=4k+1; 1 if n=4k+2; i if n=4k+3; k an integer. 5. (a) 117+44i, (b) 556, (c) 76i. 6. If and only if: (1) none of the factors is zero; (2) the factors are of the form (a+ bi) and A (b+ai), where A is a real number. 44 5i a2 b2 +i 2ab , (c) 7. (a) cos 2a+ i sin 2a, (b) 318 a2+62 a2+ ba 1 -321 , (e) 2. 25 (d) 8. 2in -2. 9. (a) x=1 +i, y=i; (b) x=2+i, y=2i; (c) x=3-11i, y= 3-9i, z= 1 7i. 10. (a) 1 . V3 (b) 1. 2 ' 2 11. (a) a2+1)2+c2 (ab+bc+ac); (b) a3+b3; (c) 2(a3+63+c3)-3(a2b+a2c+b2a+ b2c+c2a+c2b)+12abc; (d) a2 ab+b2. 1 i V3 2 , 1 i ; (b) 0, 1, i, 1, 12. (a) 0, I, + 2 2 2 15. (a) (1+i); (b) (2-2i); (c) (2i); (d) (1 +4i); (e) (1 21); (f) (5 +61); (g) . (1 +3i); (h) (1 3i);
169
(i) (3-i); (i) (3+i); (k)
( 1/V13 +2 2
i VV1-2
2
(1) V 8+21/ 17 i V -8+21/17;
(m) (/4
+
/i) ;
= 0. 1, 2, 3.
1/2(1+i)
(n)
, (o) (c' 2 - '
1-Y3
2 ()'
16. (1 3-0c0. 17. (a) x1=3-i, x2 = -1+2i; (b) x1=2 + i, x2 =1 -3i; (c) x1=1- t, .7c2 5 18. (a) 1 2i, -421,
4-2i
(x2- 2x + 5) (x2 + 8x+ 20);
(b) 2 1 1/2 , -22i1/2, (x2 -4x+6) (x2 +4x+12). 19. (a) x= 1r7 i (b) 4i. 2 2
V9 P
22. (a) cos 0+i sin 0; (b) cos t:+ i sin TC (c) cos (d) cos arc
7
2
i sin i
2'
+i sin
: +1 sin 4 7 ,4 I ; 7(e) 1/ 2 (cos 1

(g) 1/ 2 (COS
(f) vy (cos .37` 4 +i sin 3n);
+i sin 4) ;
741 7t+isin 77' 4-); (i) 2 (cos -(cos (h) 1/2
+i sin 1 7 3) ;
2rc 4-rc +1 sin (j) 2 (cos 3 - +i sin sin -0; (k) 2 (cos (1) 2 (cos 3 +i sin -d; (m) 2 (n) 3 (cos
(cos 2- i sin i 7)
. 11 rc 7C+ i sin >z); (o) 2 (cos 117r +i sin 6 6
tf +i sin (p) (1/24 V6) (cos 2 12 ) Remark. Given here is one of the possible values of the argument. 23. (a) 1/10 (cos 1826'+i sin 1826'); (b) 1/ .(cos 34557'48"+i sin 34557'48");
170
(c) V 5 (cos 15326'6"+i sin 15326'6"); (d) 1/-5 (cos 24326'6" + i sin 24326'6"). 24. (a) A circle of radius 1 with centre at the origin. TC to the positive direction (b) A ray issuing from the origin at an angle of 6 of the axis of reals. 25. (a) The interior of a circle of radius 2 with centre at the coordinate origin. (b) The interior and contour of a circle of radius 1 with centre at the point (0, 1). (c) The interior of a circle of radius 1 with centre at the point (I, 1). 3 3 26. (a) x= 2i, (b) x= +i. 2 4 27. The identity expresses a familiar theorem of geometry: the sum of the squares of the diagonals of a parallelogram is equal to the sum of the squares of its sides. 29. If the difference of the arguments of these numbers is equal to n+ 2kn, where k is an integer. 30. If the difference of the arguments of these numbers is equal to 2kn, where k is an integer. 34. cos (9+ cP)+i sin (9+ 4)). 35
2 [cos (297 2 1 )+i sin (2<p

(c) (2 1/ 3 )12, (d) 64.
36. (a) 2'2(1 +i), (b) nn 38. cos +1 sin . 3 39. 2 cos 23" 40. Solution. 1+ cos a+ i sin a =2
a 2 cosy - + 2i sin
cos -f =2 cos i (cos 2 +i sin 2 );
(1 +cos a + i sin a)" = 2n cos" i c 43. (a) i, 1/ 3 +i 2 '
(cos
+ i sin 2I .
1/ 3 +i
2 '
(b) 1+i, 1+1/ 5-.4_ 1/3 -I i 2 ' 2 ' (c) 1+i, 1 i, 1+i, 1 i;
I -1/ 3
2
1 +1/
2
.
1;
(d) 1, 1
(e) 11/3,
1 4
2
la
2 '
1 . V3
2
1+. 2 ' -2 +1
VT
2
1 . 2 ' 2 1 2
3+iia-3+fra
2 ' 2 ' 2
CH. I. COMPLEX NUMBERS 6
171
44. (a) where

6
ck= cos
(cos 85'18"+i sin 85'181 ek, 120k +i sin 120k, k=0, 1, 2;
(b) 1/10 (cos 11351'20" + i sin 11351'20") ek, where ek =cos 120k+ i sin 120k, k=0, 1, 2;
10 (c) 1/13 (cos
1115'29"+ i sin 1115'29") ek, where ek = cos 72k+ i sin 72k, k=0, 1, 2, 3, 4. 1 (cos 24k + 19 n+i sin 24k + 19 45. (a) 12 72 72 V2 where k =0, 1, 2, 3, 4, 5; 1 ( os 24k+5 24k +5 \ , (b) 16-is n+ i sin c 96 ) 96 1/2 where k=0, 1, 2, 3, 4, 5, 6, 7; / 24k + 17 24k + 17 7c), (c) cos n +i sin 12i 72 72 -1/ where k=0, 1, 2, 3, 4, 5.
2k7c \ where k= 0, 1, 2, .. ., n- 1. +i sin nr n 47. (a) Solution. Consider (cos x+i sin x)s. By De Moivre's formula, (cos x+i sin x)6=cos 5x+ i sin 5x.
46. f3 (cos
2krc
On the other hand, (cos x+ i sin x)5 =cos6 x+5i cossx sin x-10 cossx sins x-10 i cos2 x sins x+ 5 cos x sins x+ i sins x= (cossx-10cossx sins x + 5 cos x sins x) +1(5 cossx sin x-10 cossx sins x+ sing x). Comparing the results, we have cos 5x=cos6x- 10 cosi' x sins x+5 cos x sinsx; (b) coss x-28 cosex sins x+ 70 cos4 x sin4x-28 cog x sing x + sins x; (c) 6 cossx sin x-20 cossx sins x +6 cos x sinsx; (d) 7 cos' x sin x-35 cossx sinsx+ 21 cog x sins x-sin'x. 48. 2 (3 tan cp - 10 tans + 3 tang9) 1-15 tans 9+15 tans 9 - tan6cp 49. cos nx=cosn x-qCOSn-2 x sins x+ cosn-x sinsx - +M where
M=(-1) 2sinnx if n is even, and

n-1 M= (-1) 2 n cos x sine-lx if n is odd.
172
sin nx=Cn i cosn-1x sin x Cn s cosn - 2 X sin' x+

n -2
+M
where M=(-1) 2 n cos x sine-1x if n is even, and

n-1
M= (- 1) 2 Sirin x if n is odd. 50. (a) Solution. Let a=cos x+ i sin x. Then a".= cos x-i sin x; ak =cos kx+ i sin kx; a-k =COS kx-i sin kx.
k -a k ; sin kx - a 2 2 a+ a" a a' ; sin x= In particular, cos x2 2i ' (a3-a-3)- 3 (a-a-1) 0C-OL -1 ) 3 C(. 3 -3a-F3a -1- a -3 sin3 x = 2i -8i -8i 2i sin 3x- 6i sin x 3 sin x - sin 3x sin3 x = 4 -81 cos 4x-4 cos 2x+ 3 ; (c)cos 5x +5 cos 3x+10 cos x ; (b) 8 16 (d) cos 6x+6 cos 4x+ 15 cos 2x +10 32
Whence we have cos kx =
,k
52. Solution. (m -2p + I) P! (m-p -1) ... (m -2p +1) (p -1)! m (m-p -1) (m-p - 2) . . . (m - 2p + 1) P! Denote 2 cos mx= Sni; 2 cos x= a. Then the equation that interests us may be written thus: 1 _3) am -4 S.=an'- mam -2+ (qt....2 + Cm Cgt-p+Cfn ip-i= (m
+ ( -1)P (Cfn-p+Crj-p-1) arn -2P - p) (m - p -1)
It is easy to show that 2 cos mx=2 cos x 2 cos (m-1)x-2 cos (m -2)x or, in our notations, Sm=aSm_i-Sm- 20 It can readily be verified that for m=1 and m=2, the equation being proved holds true. Let us assume that S._1= am-- (m - 1) am-3 + (Cm 2_3+ C;r2 _ 4) am- 5
, _ 2) am--2p-1 - 1)P (Cgi _p I + cf7,7,1
CH. 1 COMPLEX NUMBERS
173
Sm_ 2 = a"' 2 (M 2)dn 4 (Cm 2 _4 + _ 5) am 8
n2p 2) am -2P + . - i C r (Cfn-1C

Then S.= a'n - ma"' 2
. . . ( 1)P (Cfn p _1 +
+ 07 ' 11P 1+ Cg?p 2) am-2P +
Bearing in mind that Cf ;,=Ca k _1+Ci n`l7.1 1 ., we get the required result. sin mx 53. si
Sill nX
(2 cos x)m-1- C, 1 7,_ 2 (2 cos x)n -8 + (-1)P Cg, _p_.1 (2 cos x)m-2P-1 +
nn
+ _ a(2 cos x)" 1-5 -
54. (a) 2 2 cos I ? C , (b) 2 2 sin 59. (a) Solution.
56.
2n nrc n -I sin 6
3 2
S=1+ a cos p+ a cos 2cp+ . . . +ak cos kcp. Form T= a sin cp + a2sin 2cp+ . . . + ak sin ky; S + Ti = 1 + a (cos cp +i sin + a2(cos 2cp i sin 2cp) + . . . + ak (cos kcp + i sin kcp). Setting a= cos 9+i sin cp, we have S + Ti = 1+ acc a9 ak
ak a k-Fi,k+1_1
aa - I
S is equal to the real part of the sum obtained. We have

ak-Fi
,k+i_i
S-1-Ti =
aa - I
a,-1 - I ock_ ak-F1 ockas-1+1 =ak+ 2 . a2 -a(a+a -')+1 as-1-1
Whence S= (b)
ak +2 cos kcp -ak +1cos (k +1) cp- a coscp+ I a2 -2a cos cp +1
ak + 2 sin(cp+ kh)._ ak +1 sin [9+ (k +1) h] -a sin (cp - h)+sin p a2 - 2a cos h sin 2n +1 2
(c)
2 sin x 60. Solution. T=sin x+sin 2x+ ...+sin nx; S= cos x+cos 2x+ ...+cos nx.
174
x x Let a.= cos 2- +i sin

1
Then S + Ti= oc2 + (ce _
...
Man,
S+Ti=ccz
2 c o
_=
n)
n+1
2
= I cos nx
x+i sin
n+1
2
\ ) x
sin
12
2x sin 2
n +1 sin
Whence T=sin
2 x
x sin 2
61. 2 (2cos x) 5-4 cos x 64. (a) sin (a + n 1 h) sin 2 . 2 , ns i even, h cos nh 1 h)cos 2 , if n 2 is odd; h cos 2
n cos (a +
cos \a+ (b)
h)
cos 2
nh sin 2 if n is even ,
sin (a +
nh n 1 h\ cos 2 ) 2 cos y
if n is odd.
66. (a) 2" cos" 67. (a) 2" sin"
cos
n+ n+2 x; (b) 2" ens" x sin 2x. 2 2 2
cos nn
(n + 2) x ; (b) -,
2
x (n + 2)x nrc 2" sin" 2 sin 2
68. The limit of the sum is equal to the vector depicting the number 3+i 5 sin 4nx 69. 2 4 sin 2x
175
3 cos
n +1
2 4 sin
n x sin x n
cos + sin
x ,` nx x si 2
3(n + 1) x sin 2 3x 4 sin -2-
3 nx
2 '
3 sin
n+ 1
3 (n + 1)
2
nx x n3
2
4 sin 2
3x 4 sin 2
(n+ 1) cos nx-n cos (n +1) x-1

2 (n+ 1) sin nx - n sin (n + 1)x 4 sins 73, ea (cos b + i sin b).
75. (a) -3, 3+i - -1/ 3 x
4 sine
(b)
(b) -3,
3+54 3
2 55i
3 3
(c) -7, -1+4 3 ; (d) -1,

3 3
(e) 2, -1+1/S-
(f) -1/ 2 - -1/4 , 1/71.-1/2 (g) -2-1/3
11/ 3
2
( -1/T+ 1/fl;
21/32
'1/9 +i1/3 (1/9 + 2 1/3); -
(h) 1 - -1/ (i)

3 3
2+ -I/ 2 +-1/ 4 2
1/3 2
1 (-1/7
-1/;
-2-1-1a+TrT
2 - 2 ; (k) 2, -1 34 3 ; (1) 2, -1+41 -1/ 3 ; (j) 2, -12i (m) 1, -21/3; (n) 4, -1 4i-ra ; (o) -24 1, i; (p) -1-i, -1-i, 2+2i;
(q) - (a+ b), a+ 2 b +1 23 (a- b);
3
(r) -
Vf 2 g+bil fg9,
3 3 3
allig+171/ fg2
2
/ 3 (a-1/ fag-b e fg a); 2
176
(s) 2, 1149, 0,2541, I, 8608; (t) 1,5981, 0,5115, 2,1007.
76. Solution. xi x2= oc (1 co) + (3 (1 co2)= (1 co) (a (30); xi x3=a (1 co2) f3 (1 co)= (1co2) (a (30; x2 xa=a (co co2)+ (3 (cot co)= (co co2) (a (3);
(xix2) (x1x3) (x2x3)=3(co 6.12) (00-133); (x1x2)2 (x1x3)2 (x2 x3)2 = 27[00+( 3)2-400(31= 27q2-4p3.
77. Solution.
The cubic equation mentioned in "Hints to Solutions" z2-3(px+ q)z+ + x3+1133qx 3pq = 0 having the obvious root z= (x+p). The other roots x+p-1/ 3 (xp)2 + 128By virtue of Problem of this equation are z2,3= 76, the left member of the equation under study may be given as 1 (z5 z3)2 (z3z1)2(z1z2)2 27
[ -3;(x p)2 + 128] [ 3 (x +P)+ V x
2 3 (x p)2 12q 1
2
[ 3 (x+P) V 3 (xp)a+12q 2 = [(xp)2-4q] (x2+Px+pa q) whence the roots are readily found: x1,2 =p2 -1/q, x3 x4
X5 = X5 =
p+4q 3p2
2
p 1/ 4q-3p2
2
78. The left member will be represented in the form ce6+135+5(a+ (3) (.2+4 + (32 a) (a(3 a) 2b =0. Answer. x=a+ (3 where
5 5
/ b+-1/ 1.2a3,
p=1/b
b2 a5; aft= a.
79. (a) VI; 1 i VT; (b) 1 VT; i V 3 ; (c) -1/2, 1i1/ 3 2 (e)
(d)
I -I V 5 2
+
2
1 T3 ; 1 i; (f) 1 -" ; 2 2
5+i lrf
2
177
(h) -1-175 1 i1r7 2 (i) i, 10/6;(j) 2+21/2, 1+4 (k)1, 3, 11/ ; (1)1, 1,1+2i; (ln) 1+1/22+21/5 1-1/5+.1/22-21/5 4 4 (g) +i, (n)
(0
1+1-/51/30-61/ 5
4
1V51/30+6Vg
4
1+V2 I -,/
2
'
V I 21/2,
1V
2 I V 1 +21/1;
(p) I +1/ 7- 1/6+21/7, IV7 +V6-21/7 ;

(c)1V41/3
1V 4V3-3 ,
2
(r) 1+1/5V-2 61/5

4
1-1/5V-2 +61/5
4 1/j/-5-21/2 . 4
(s) 1+1/2 V-5+21/2

4
(t) 1+1/31/12+21/3 ,
4 80. Solution. x2-faxa+bx2+cx+d = (x2+ xi X 2 +n;
1V3 V12-21a
4
x+ 2 +mx+n) (x2 + a x+ n; A=xix2+x3x4 . ; (c) 1, i;
mxn);
2
x3 x4 =
81. (a) 1; (b) 1, (d) 1, (f) 1, +i, 1
i
1 2
"
2 +i
(e)
+1 i " (1+i). '

,
1/3, 2 2 2 ;
(g) I,
2
4
, 1/22 (1 1), 2
+V2 +i 4
V6-1/2 +i 1/6+1/2 4 4
178
82. (a) -1; (b) (e) 2 (g)
1 -1, i 2 2
. (c) 1 V3 . +i , 1; (d) 2 ' 2 2 2 ; 4

i
-ii
(1 i); (f)
-1/-
4 4 83. (a) 20, 20, 180; (b) 72, 144, 12. 1C7T 2lac where k=1, 2, 3, 4, 5, 6. 84. cos +isin 27 7
,
V6--V 2 +i V 6 +V 2 1/+1/ +i ---VT , + 4
+i sin we get the following: 16 16 ' eo belongs to exponent 1, esbelongs to exponent 2, 4, C12 belong to exponent 4, ea, e6, C101 e14 belong to exponent 8, the primitive 16th roots are EEEEEE -1, -, 5, 73 9, 11, e12, 15 krc (b) Denoting Ek = cos 220 +i sin 21or , we find that 20 E. belongs to exponent 1, eio belongs to exponent 2, En e nbelong to exponent 4, en es, C12, C16 belong to exponent 5, en C6, e14, els belong to exponent 10, the primitive 20th roots are en ea, Cif C9, Cu,C13, C17, e19. 27 i sin 2 (c) Denoting Ek= cos 21 24", we find that
85. (a) Denoting ek = cos
2krc
2krc
e5belongs to exponent 1, enbelongs to exponent 2, En Enbelong to exponent 3, e6, enbelong to exponent 4, en en belong to exponent 6, E., enels, en belong to exponent 8, en m 14, e55 belong to exponent 12, the primitive 24th roots are et E-19, E if -52 -if -112 E -13, -17, -33. 86. (a) Xi (x)= x - 1 ; (b) X2 (X) = X + 1 ; (C) X3 (x)=x2 +x+ 1 ; (d) /4 (X)=X2 +1; (e) X5 (x)=x4+x2 +x2+x+1; (f) /6 (X)=X2--- X+ 1; (g) X, (x)=x6+x6+x4+x3+x2 +x+ 1; (h) A'. (x)=x4+ 1; (i) X. (x)=x2 +x2 + 1; (j) Xi (x)=x4-x2 +x2-x+1; (k) (x)=x10+x9+x8+x2+x8+x5+x4+x3+x2+x+1;
,
CH. 1. COMPLEX NUMBERS (1) X12 (X)=X 4 X2 + 1 ; (m) Xls (X)=X8 X7 X5 X4 +X3 X + 1 ; (n) X105 (x) =x48 4_ x47 x46 x43 x48 2x41 x 40 x89 + x86 +x85 + x84 +x83 + x82 +x31
X24
179
x28
x28
--)C22 X80 +X" +X16 +X" +X" X13 X12 x9 X8
2.e .X6 X5 +X2 +X+ 1.
87. 1 c 88.0 if n> 1. 89. n if k is divisible by n; 0 if k is not divisible by n.
90. m(x'n+1).
91. 11 lc if e01; 0,
n+ 1) if
2
a=1.
92.
n2(1 + 2n
0_
if col;
n (n+ 1 )(2n + 1) if c=1.

6
v 93. (a) i ; (b) i cot n 94. (a) 1, (b) 0, (c) 1. 95. x0=1; xi= cos 27c
+i sin
2n V 5 1 i ./ = + 4 V10+21/5 ; 4 5 1/. +1 i + I 10-2 V 5 ; 4 4 'V
x2 =cos 5 +i sin 4Tc 5 = x3 = cos x4 = cos
. 6n 6n V 5 +1 + i sm = 5 4 5 8n 8n -1 +i sin = 4 5 5 1 ' cos 18=
i,1 V 10 2 1/5; i -,/ 4 v 10+2 V 5 .
96. sin 18 Vg 4
V10 + 21/5 4
97. Solution. Divide both sides of the equation x6 +x5 +x4+ x2+x2+x + 1= =0 by e. A few simple manipulations yield 2 1 3 -) 2 (x+ c I ) 1=0. (X+ + (X+ -I x
The equation z2+z2-2z 1 =0 is satisfied by z=2 cos 4 " = 2 sin 7 14 satisfies the equation t3 t5-2t+ 1 =0. The resulting Whence t= 2 sin 14 equation is the simplest in the sense that any other equation with rational coefficients having a common root with it is of higher degree. The proof of this fact requires information from subsequent sections of the course.
180
1 and -1 and 2m-2 complex roots. Here ek =cos sociated with e2,_ k=cos
xvn
-
98. Solution. Let n=2m, then the equation xn - 1 =0 has two real roots, . 2kir is as+i sin
2m
sin + i sin
(x C2)
2m
We thus have
1 = (.0 - I) (x - s1) (x - Ez.) (x - ea)
(X CM -1) (X C M -1);
(ei +El) ' - 1) [X2 X2M 1 = (X m-1 x2m

-
x+ I] [x2- (Em-i+
n. +1).
1=(x2
1)
k=
fl (x2 -2x cos

k =1
If n= 2m +1, then in analogous fashion we obtain . 2 - 2x cos 2kTr +1) x2.+1_1=(x_1) n (x 2m+1
99. Solution. (a) We have m- I
Xx2271 =
n (x2- 2x cos : +1)

k =1
m-i Putting x=1, we get m=2"2-1

k =1 mI
(l _cos
krc \ nz
or m=22 (m --1-)
k =1
sine
2m
and finally,
m_i
2m-i
vm _n sin
k=I
krc
2m
Formula (b) is obtained in similar fashion.

n-1
100. Solution. In the identity xn 1 = n
(x_.,)
a . We get _
k =0
where ek= COS
2kir
+i
sin an , put x=
n
an (- 1)n - 1 = (- on bn
k) and so on. ;e n (1
k=0
181
101. Following the suggestion made in "Hints to Solutions", we get

n-1
cos ne + i sin ne 1=11 (cos

k=0 n-1
i sin 0 ek),
cos nei sin ne 1=
n (cos o_i sin 0 ek).
k=o
We get the required result by multiplying together the last equations. 102. Solution.
n-1 (t ekr t 1 k=0 n-1 n 1 n--1 n,
J] H (t+SkES)
k=0 s=0 n-1 n-1
tn
k=0 s=0 n-1 n-1 1 =t o
[t k
ek
1)1
1 to k=0 s=0 n-1 1
t_ ek (es l)]
Li Li
[t ek (Es 1)]=
n[ [ tn_ (s_ On ]
s=0 n-1 [t n (e k I )1k=
s=0 k=0
103. We have I x I=Ix 1n-1, hence I x 1=0 or I x 1=1. If I x1=0, then x=0. But if I x1=1, then xx=1. On the other hand, x.Tc=xn. Hence, xn=1. Thus 2kn 21crc +isin , k=0, 1, 2, ..., n-1. x=0 and x= cos The converse is readily verified.
n
104. Solution. If z satisfies the given equation, then z a zb The locus of points the distance from which to two given points is the given ratio is a circle (a straight line in a particular case). x+1 21crc 2krc 105. (a) We have ck where ck = cos + sin x-1 k=1, 2, ..., m-1. Whence x= ek+ 1 . Transformation of the last exek 1 krc pression yields xk = i cot , k= 1, 2, ..., m-1 ; (b) xk =cot kir , k= 1, 2, ..., m-1;
182
(C) X k = n
en/ 2 1 where ek = cos

2krc 2krc +i sin n n
k= 0, I, 2, . . . , n 1.
106. Solution. Let A= cos cp+i sin cp. Then
1+ix
1 ix
=12where 71k=cos
k
y+21crr y+2/cir , k=0, 1, ... , in-1 +i sin 2m 2m
Whence,
x=
712 k 1 Yak -71k I cp.+ 2krr 2m / (74! +1) i(7)k+7V) tan
107. Solution. Using the hint, we have S+Ti=1..t (1 + Ax)a, S (1 -Joc)n
where A= cos a+ i sin ; 1.i= cos cp+ i sin cp. Whence 2S = 11(1+ Ax)a+P. (1 +)\x)11. The equation becomes (1+ x)" + 17. (1 sin
X k= =0;
(2k + 1) n
2n
2p
sin
( k + 1)rc
2n
2na
k = 0, 1, 2, ..., n-1.
108. Solution. Let ma= I, pb= 1. Then (p)4b =(aa)b . (pb)a = 1. 109. Solution. Let e be the common root of xa I and x5 I ; s is the exponent to which e belongs. Then s is a common divisor of a and b and can
therefore only be equal to 1 and e= I. The converse is obvious. 110. Solution. Let ak and psbe roots of 1 of degree a and b; k=0, 1, 2, ..., a I ; s=0, 1, 2, ..., b 1. On the basis of Problem 108 it suffices to show
k = Ps , s.. Then L that all ak P sare distinct. Suppose that al, 133, =ak. p ack. Ps, i.e., oci= pi. On the basis of Problem 109, a= pi= 1, i.e., k1= k,, s1= se. 111. Solution. Let a and p be primitive ath and bth roots of 1. Let (ap)s=1. Then ad's= 1 , pas= 1. It thus appears that bs is divisible by a and as is divisible by b. Hence, s is divisible by ab. Leta be a primitive abth root of unity. Then X-= ps (Problem 110). Let ak belong to the exponent al< a. Then V .I. = 0,19a,6 (ps)a,b = I which is impossible. In the same way it may be shown that Ps is a primitive bth root of I. 112. It follows directly from Problem 111. 113. Using the hint, write out all multiples of p that do not exceed pm. p"-1 p. It is immediately seen that there are Namely, I p, 2 p, 3 p, pc- such numbers. Whence y (p")=p"pa '=pa (1 1).On the basis
183
) (1 )
of Problem 112, cp (n) =cp (Pi')

(1P -1k
cp (pc k`k)= n (1
PI
P2
114. Solution. If e is a primitive nth root of 1, then e, the conjugate of e, is also a primitive nth root of 1. Here, e 1 since n> 2.
115. Xp(x)=xP-'+xP -2 +
116. Xpm (x)=. x(P-
. . .
+ x+ 1.
2)/In I+ xpin + 1
1) ptn I
+x (P
117. The suggestion can be utilized immediately on the basis of Problem 111. Let al, oca, a9(n) be primitive nth roots of 1. Then al, oc 04.9() are primitive 2nth roots of 1. We have X2. (x)= (x+al) (x+c(2) . . . (x+ot,p()) = ( 09(0 ( or (Problem 114) X2,, (x)= X,, (x).
118. Solution. Let ek = COs 2kn
CO . . .
MVO)
be a primitive ndth root i sin nd nd nd of 1, that is, k and n are relatively prime. Divide k by n, this yields k= nq+ r, 2ris 2qrc-F 2 rc+ 2rrc 72 where 0 < r <n. Whence ek=cos d that is, + i sin q = cos 2rrc + i sin 2rn n n ' r is a primitive nth root of 1, since every common divisor of r and n is a common divisor of k and n. 2rrc 2rrc Now let n r = COS be a primitive nth root of 1, i.e., r i sin ek is one of the values of the dth root of and n are relatively prime. 2rrc d 27c (r + nq) where + i sin
nd
2qrc + i sin
Form eq cos
2ris n
cos
2-rc (r + nq) nd
0, 1, 2, ..., d 1; eq is a primitive ndth
root of 1. Indeed, if r+ nq and nd were both divisible by some prime p, then p would divide n and r, but this is impossible. 9(,,,)be primitive roots of 1 of degree n'. 119. Solution. Let e2, e2,
cp(n')
Then X,, (xno=

... (X
fl
(x" ek). Furthermore, let

-
(x ek, i) (x ek,
k =1
ek,,,-) be a factorization of x" ekinto linear factors. Then X;, (xn") = ek, ). On the basis of Problem 118, each linear factor x ek,
Ic=-cp(n') i=n"
k=1 i=1
184
enters into the factorization of X,, (x), and conversely. Besides, since p (n)= n"cp (n'), the degrees of X,, (x) and Xn,(el are equal. 121. Solution. The sum of all nth roots of 1 is zero. Since each nth root of 1 belongs to the exponent d, which is a divisor of /1, and conversely, it follows that
E (d)=0.
din
210r 2 k rc 122. Solution. Let ek = cos +i sin belong to the exponent

n n
n1. Then the factor x ek will enter into the binomials Xd 1, where d is divisible by n1(and only such binomials). Here, if d runs through all divisors n of n that are multiples of n1, n runs through all divisors of d ni Thus, x ekwill enter the right side with exponent
d
E (.,. (di). This sum is
equal to zero if n 01and to 1 if n=n1. n, 123. Solution. If n=p'l where p is prime, then X,, (1)=p. If n=p7' PZ k, where pi, Pi, . pk are distinct primes, then (Problem 119) X,, (1)= = Xn,(1), where n' Pi Pk. Now let n=p1, Ps...Pk; 2 ; n,=. Note that in order to obtain all divisors of n it is sufficient to adjoin to all divisors of n1their products into pk. Therefore
Pk
Xp (X)=
fj (
din
xd_i) d
C I)
=
din,
(x
d_ (
d) L2 din,
(x
dp , K 1)
"Pk
= [Xi, (XV I' A 'n, (P).
Whence X (1)=1.
Xn
124. Solution. (1) Let n be odd and greater than unity. Then (Problem 117) (-1)= X2.(1) =1.
n
(2) Let n= 2k, then X n=
xn-
n
x2 I
=X
+ 1 and Xn (-1) is equal to 0
if k=1 and to 2 if k >1. (3) Let n=2n1where n1is odd and greater than unity. Then (Problem 117) Xn(-1)=Xn (1) and, hence, Xn (-1) is equal to p if ni=px (p prime) or is equal to 1 if n1 0p'. (4) Let n = 2k n1, where k >1 and ni =PT' Pc are 2 '2 /4s (P1, P29 distinct odd primes). In this case (Problem 119) Xn (x)= X2pp, ps (x'), where A=2k 1 pT 1.. pis I . Whence it follows that X ( 1)= X,, (1) = 1.
185
125. Solution. Let el, e2, , e,p(n) be primitive nth roots of 1: rt-t (n)12 (e+ + e q) (17) ) s =e1 e2 + e3+ . + e<p(n) 1 e<p(n)= 2 (1) Let n be odd, then c7 is a primitive nth root of 1 and e7= e3 for i=1 only. Therefore !-E
2 (n) = (n) and + . . . + e,p
s=
(n)I2
(n)
(2) Let n=2n1, n1is odd. In this case ei(Problem 111) is a primitive nith root of unity and therefore [see (1)] +...+ )= 0'0= (n).
(n
[P. (n)12 + (n) Thus, in this case s

2 2 -2 . (3) Let n= 2kn1where k> 1, n1is odd. In this case ea belongs to exponent 1 On the basis of Problem 118, we assert that el, 2, , eq, (n) are square roots of rhoh, where x)1,112, G) of 1. Whence it follows that n are primitive )1(P G) th roots
+ + . . . + e9 2Kn) =2(7)1+712 + +
9
S = tot
(0)
y) ;
n 1
Y n 1 x= y
n-1
n 1 s =0
n-1
126. Solution. S= E cx.= E ex.= E e( y+s).
x=0
for arbitrary integral y:
n-1 n-1
E e Y% S' S= E .Ys= E (zy.. E .(y+s).)

y=0 y =0
n-1
y =0
n-1
s= -0 n-1
n n-1-1
=
2ys -Fs = Ee
E ( es' E c,,,$) = n+
cs.
y =0 s = 0 s =0 y =0 s = 1
.E
n-1 ( e2sN y ) =n
for n odd;
n)2
1+ (-1) 1 SS' =n+nek21 =n [

n-1
for n even (since E 2sy=0 for 2s not divisible by n).

y =0 To summarize, I S I = 1/n if n is odd, and I S if n is even.
186

(e) a2 p2 y2_ a2, (f) sin (a(3), 127. (a) 5, (b) 5, (c) 1, (d) ab c2 (h) sect a, (i) 2, (j) 0, Zk) (b c) (d a), (1) 4ab, (m) 1, (g) cos (a + 1 +i2 3 (n) 1, (o)
128. (a) 1, (b) 2, (c) 2a2(a+ x), (d) 1, (e) 2, (f) 2 --V 2 , (g) 3i 1/ 3 , (h) 3. 129. The number of transpositions is odd. 130. (a) 10, (b) 18, (c) 36. 131. (a) i=8, k=3, (b) i=3, k=6. 132. q. 133. 135. (a) 134. (a) 2 , (b) r
n (n-1)
2
, (b)
n (n 1)
2
n (3n+ 1)
3n (n 1)
2
136. Consider a pair of elements ai and ak, where i <k." If these elements do not form an inversion, then ai will precede ak when the permutation returns to its original order, and, hence, the indices will not yield an inversion. But if the elements a1 and ak form an inversion, then akwill precede ai after the permutation is returned to its original ordering and, thus, the indices i and k will yield an inversion. 137. In both cases the permutation is odd. This is due to the fact that the original arrangement is obtained from the other one by means of an even number of transpositions. 138. (a) With the + sign; (b) with the + sign. 139. (a) No, (b) yes. 140. i= 1, k=4. 141. a11a23a3sa44, a12a20/34a41 anda14a23a31a4a a31 a32 as6 142. aua2c1 a41 a42 a45 a6i
a5a
a66
143. With the+ sign. 144. With the sign ( OC72 7.

n (n-1)
146. 2, 1. 147. (a) n!, (b) ( 1)

n (n1)
, (c) 2
nI.
n (n+1)
148. (a) (-1)
(n !)n+1
(b) (_ 1)
oon-I-1.
149. Solution. Interchanging rows and columns : (1) does not change the determinant; (2) makes the determinant a conjugate complex number. 150. Solution. Interchanging rows and columns: (1) does not change the determinant; (2) results in the determinant being multiplied by 1.
n (n-1)
151. ( A. 152. Is multiplied by ( I) 2 153. Zero, since the number of even permutations of n elements is equal to the number of their odd permutations.
187
154. (a) xi= ai; x2= a2; ...; xn_ = an-1; (b) x1=0; x2 =1; ...; xn _ 2= n-2; (c) xi= ai; x2 = a2 ; ...; xn - = an -1. 156. 0. 158. (mq_np) la GI
Ic d (-+ --+ . . . + -) a]. an az

ai a2
159. (a) ai az . . (b) ( -1) 2
n (n-1)
a,, I 1 \ a1
1 a2
1\
an
160. 3a- b + 2c + d. 161. 4t-x-y-z. 162. 2a- b- c- d. 163. -1,487,600. 164. -29,400,000. 165. 48. 166. 1. 167. 160. 168. 12. 169. 900. 170. 394. 171. 665. 172. a' + b' + c2 -2(bc+ ca+ ab). 173. -2(x3+3,2). 174. (x + 1) (x2- x +1)2. 175. x2z2. 176. -3(x2 -1) (x2-4). 177. sin (c- a) sin (c - b) sin (a- b). 178. (af- be + cd)2. 179. n!. 180. bib 2...bn. 181. (x- xi) (x - x2) .(x- xn). 182. (n - 1)!. 183. - 2(n- 2)!. 184. 1. na"-1 [2a + (n - 1) h]. 186. 185. 2 [2a + (n - I) h]. 2
n (n+1)
[ao - ai+az - . . . + ( - 1 )" an] x" - 1 nx" 188. ao xn+ ai x"- + . . . + a 189. (x- 1)2 ' x- I 190. (n+1)! x". 191. (x-a,) (x- a2). .(x-an).
187. (-1) 2 192. [x+ (n-1) a] (x - a)"-1. 194. (-1)" (n+ 1) alga. an 195. ai az 196. h (x+h)n. 198. (199. (-1) 193.
(x+ a)n+(x2
+...+7 1 1 4-+ (1, ) . (1+ 7

197. (_i)" -1
_
4..
xn- 2. 1 a2 + 1\ an
(n+ 0'1+1
nn
2" -1ai a2 . an
n (n-1) 2
1
a1
nn-1(n+1)
2 202.
200.
201.
11 ( 1 - a x , k).
k=1
2n-2 (n+ 1).
188
203. (-1)" (clob0+ aibi+ ...+ anbn)
204. a (a+ b) (a+2b) ... [a+ (n+ 1) b].

205. xn+ (- 1)" -1 yn. 206. 0 if n> 2. 207. 0 if n> 2. 208. Let n=2. Using the hint, we have I
ai + x2 ai+ a2+ xi1+ a2+ xa 1 0 01 1 ai+ x,

0 a2+ x, + a1+ x2 a,+ x, a8+ =1 + [(ai+ xi) + (a2+ x2)]+ (a, ai ) (x1 x2)
ai+ xi 0 a2+ xi 1
If in this fashion we represent an nth-order determinant in the form of a sum of 2" determinants, we find that one of the summands is equal to unity, n (n 1) summands n summands are equal to ai+ xi, where i=1, 2, ..., n and 2 are equal to (ai ak) (xk xi), where i> k. The other summands are zero. Thus, we have the answer:
I+
E (ai+ xi) + E (a;

i=1 i> k
ak)(xk xi).
This result can be transformed to (1 +ai+ an + ...+an) (1 +xl-Fx2+ ...+xn)n (aix1+a2x2+ ...+ anxn). n +1 209. 0 if p> 2. 211. 1x xn+ 1 (1.702
, 212. Solution. It is easy to see that ....2 =xix2(1 + +...+ =xix2. ..xn _ i(1 + xi X2 An.- 1
x2
a,
x1
=). Suppose that

X2
a,
an_
al a, ) A- + +an-1 Then An =xi x9 ...xn (1+ xn _ 1 X1,
. ai +a, +...+ + an xi x2. .xn _i= x,. xn (i + = xi X2 Xn

213. aox1x2...xn+aiyix2...xn+a2Yvax3...xn++anYILY2. Yn
214. an
) ( + + . . . + an an . ai an
Onal+ ( 1)n+ I.
215. n! (anxn + aixn + . . .+ an). 216. aia,...an_ 1 aia,...an_ 2+

this yields
217. Solution. Expand the determinant by elements of the first column;

An = (cc +P.) An-1 c(P 0n-2- It is easy to verify that
CH. 2. EVALUATION OF DETERMINANTS pa p3 A2 = as Aa = Da- (3 '
189
co pn
p4 co -1 _ pn -1 . Suppose that An-2= p -13

co_ pn
CCP
oc - p
. Then a n 1_ pn 1 co -I- 1_ pn + 1
a -p
- 13
Alternative solution. Represent An as a sum d,,+8,,, where a aP 0 ... 0 1 a+ p ari . . 0 a+13 ... 0 d= 0 1 0 0 0 (3 0 0 1 ix+p a (3 1 a+ P n= 0 0 0 0 ... ... ... ...
0 0 0
1 a+ P 0 0 0 0 0 0
... 1 cc + p
Take a out of the first row of doand then subtract the fi ocr2s .t r ifetd wfn r_om i=cen the se1 cond. This yields dn=adn_ i. It is readily seen that da= then d=acn. Expanding anby elements of the first row, we see that an= PAnFrom the foregoing it follows that An = an+ R r An _ i. We can readily check that
,s_ t33
(Do
A2= Suppose
a- p
that An _1=
cofin
04 p
. Then a- (3
co+1_pri+1
Lin= an +13
a- p
sin (n + 1) 0 218. n+ 1. 219. . 220. cos ne. sin 0 t _1 x n 2 + cn 2 2x n4 221. JO- Cn ... Compare with Problem 53.
ni
222. xl y
(Xi +iYi xiYi
223. a2. an (I+

n (n-1)
a2,
1 +...+ 1 1
\ 1 an
an
1
224. (- 1)
4ai aa . .an (1+ a1 + as

1 1
1
+
1
225. x (a,- x). .(a,,- x) x +
x+
+ an _ x ) .
190
226. (x1-a1) (x 2- a2) .(xn - an) (1+ x 1 - a, + . . . +
an
227. ri ( xi_
j=1
a b i) (1 +
I
i =I
bi)
228. (-1)"m" (1-
_x m . E i=i
al
C4i
229. xi=x2 =...=xn _1=0; xn= 230. (a2 - b2)".
1 1 231. a (a + b) .[a + (n - 1) b] ( a + a+b 232. xn-1

1=1
n
1
a+(n-1)b) '
(x -2ai) (x +
i=1
n
\ 2ai )
(x-2ai)(x+ E x -2a ) ' i =1 =1 234. Nth bib,b3+ ...+( On b1b2...bn. 235. (-1)"-1 (bia2a....an +bibaaa...an + 236. (-1)n-1xn-2. 237. (- I)" [(x- 1)" X"].
233. xn-1
-
[1
n
i=1 m (m+1) CA++In C;;,+in -1 Ci4. nn+1n-k+1 . 244. (- 1) 2 243. CkiVI CklVi_ ... C;24-1
245. (x - 1)n . 246. (n-1)! (n-2)!...1! (x- On . 247. an.
238. aoxn J] (bi- ai). 240. 1. 241. 1. 242. 1.
248. Using the hint, we have = (x z) An (x y)n

-
An = (x-y) An _ 1-Fy(x-z)"-1. From the resulting system of equations, we find

y)" - y (x-z)" An - z (x z- y
n (n +I)
249. (-1)
ab (bn 1
-1)
a- b
191
250.
xf (Y)Yf (x)
x -Y
where
f (x)= fl
k=1
(ak x).
251.
f (b) (a) b b
-
where
f (x)=
k=1
(ck-x).
252. (oe- on- 2 [Au+ (n-2) a(3 -(n-1) al)].

n (n-1)
253. (- I)
2
n (n-1)
n"-(n+I)
2
254. (-1)
(nh)"
255. (1-xn)n -1.
256. If all columns are added to the first column, then we can take a +b+
+c+d outside the sign of the determinant; all elements of the remaining
determinant will be integral expressions in a. This proves that the determinant is divisible by a+ b+ c+ d. If we addthe second column to the first and subtract the third and fourth, we find that the determinant is divisible by a+ b-c- d. Reasoning in this way, we will show that the determinant is divisible by a- b+ c-d and a- b- c+ d. From the foregoing it follows that the determinant is equal to (a + b + c + d) (a + b c - d) (a- b + c- d) (a- b- c + d). To determine A, note that the coeffi cient of a' must be equal to 1, and so A= 1.
257. (a+b+c+d+e+f+g+h) (a+b+c+d-e-f-g-h)x x(a+b-c-d+e+f-g-h) (a+b-c-d-e-f+g+h) (a-b+c -d+e-f+g-h) (a-b+c-d-e+f-g+h) (a-b-c+d+e-f -g+h) (a-b-c+ d-e+f+g-h).
258. (x+ al+ a2+ ...+ a.) (x- al) (x a2). .(x - an).
n (n-1)
259.2
2
1-41<k, -.5n
sin
n2i>k>1
Pi+Plc 2
1<i<kn pi-ppk
sin
9k -91
'
n (n-1)
260.2
I-I
cos
2
n_?-i>k?.--1 (ai ak)
sin
cpipk
'
261. 11 2! .. .n! 262.
ri
n+1.-k>i,.1
263. (-1)" 11 2!...nl.

264. (-1r
11
1=1
(ai
ak)
1=1
ai f ' w (iai))
102
PART W. ANSWERS AND SOLUTIONS
where f (x)=(x -ai) (xa2)...(x--an).
265. J] (xi_xio
n (n-1)
n
n-i>k>1
266. 2
cos
fl
n.?-i>k>1
sin
-Tk
267. fl
n>i>k>1 n (n-1)
xo
an C102 ao n-i
268. _ 9
sin
cpi+cpk
sin
n>i>k>1
269. 1! 21... (n- 1)!
n (xi
n
Xk).
271. 11 3! 51...(2n-1)! 272. 1 1 i= I 273.

(bkai-akbi).
Xi xi -1 (xi -Xk).
274.
n
1<i<kn
sin (xi- ak).
275.
1-i<kn-F1
276. 2(n - 1)2
(ai- ak) (a, ak - 1).
n n-I>i>k>0
sin
pi-Fcpk
sin
(1)k
277. 2n (n +1) sin (xo sin al

...sin an
sin
n
(xi -Xk)
sin
ai -ak
278. [xi x2 .. . x,, (x1 1).. (xn - 1 )1

-
279. x1 x2 . x0
xl
x2
4- "
\
(xi-k).
(XiXk).
280. (xi+ xa + . . . +xn) 281.

an s
11 (xi-xdwhere crpforms a sum of all possible pro-
ducts of the numbers x1, x2, ..., xn taken p at a time.
193
Xk).
282. [2x1xa .
)Cn 1 )] (x1-1) (x2 1)(
11 (xin>i>k>1
283. x2 (x2- 04. 284.2x8 y

n (nI)
y). (y x)n. (y x)k (y2
285. 11 21 31...(n-1)1 x 286. 11 21 31. . .(k
2
k (ki)
1)1 x
.x)k
(Y n - k
X)k
11 cy,
d2),
yj).
287. (yx)k ("). 288. (b) 9, (c) 5, (e) 128, (f) (alasbi b2) (ci ca (g) (x3 x2)2 (x3x1)2 (x2 x1)2, (On [oc + (n 1) [3], (h) (A2 a2)
(k) (x4x3) [(x3x2) (x 4 x 2)-2 (Xs xi) (x4 x1)), (m) 27(a+ 2)3 (a-1)6 [3 (a + 2)2 4x9 [3 (a 1)24x93. Remark. This problem is a particular case of Problem 537. 289. (a) 7 1 4 4 5 8 5 5 4 4 2 4 3 3 5 0 , (b) 3 3 4 3 2 11 3 8 6 8 17 5 3
(c)
290. (a) 24, (b) 18,
(c) (a+b+c+d) (a+bcd) (ab+cd) (abc+d).
291. (a) 256, (b) 78400, (c) (aa+b2 +c2 +d2)4.
292. D
293. (a) C, I , a... Cl,"
n
n>i>k>0
(al a k) (b
bi),
(b)
ni>k1
294. 0 if
2. 295.
ri
i=1 n?-i>k1
296. (a'+ b2 + 0+,12+12+ m2 +,2 pa)4. 297. 4 sin4 cp. 298. 4 sin4cp.
7. 1215
194
PART 111. ANSWERS AND SOLUTIONS
299. Denote the desired determinant by A. Squaring yields IA I=n 2 (Ek_ es). On the other hand, A= 11 We assume ei=cos A=
7r
+i sin
TC
.
Then z= el and
(ck_ o
= =
elic+s
{
n (n-1)
Es)
e,
k+s
n
(k
2 sin
(k s)
n
Furthermore, sin
n
(k s) rc
>0 for all k, s. Hence

(k s) TC n
n=I A 1= Therefore
is n (n 7 1) A= n 2 i 2
n 2 sin
= n 2 sin
n n (nI)
ns
n (n-1)'
k-l-s=n2 i 2 et
e, 2
+(n 1)* n (n-1)(n+2)
is n (n-1)
=n 2 i
n-1
=n2 i
300. fl
k =0
, (aoai ek + az 6/2
ary _1 4-1)
where ck = cos
2kis
n 2krc
+ i sin 301. x4 y4+z4 u4 + 4xy2z+4xzu2 4x2yu-4yz2u-2x2z2+2y2u2. 303. 2n-1if n is odd, 0 if n is even. l]n n" an (n+1) [(n+ 1) 304. ( 1)n (1 an)a

305.
( 1)n-1 (n 1) 2krc
n 1
(a, + az ek +
+an er j.) where
k=0
ek=COS
+i sin
21c7c
306. cpo(t)cp, (t). .
(t) where
Pk (1)=
ektr
e k = cos
2krc 2kir + i sin -n n
195
According to the result of Problem 102, the answer can be represented as

n- 1
un_(.,c on]
k =I
307. (-2)n-1(n- 2p) if (n, p)= I, 0 if (n,

308. 2n -2 (cosn -- 1 n
Al.
309. 2"- 2 Sinn 2
[sinn (n+22)
sin'
n20 ]
22)h )] 1 -cosn (a+ (n 310. (-1)n 2n -2sinn - 2 4 1 [cosh (a+ 1) 311. (-1)n-'
n-i
(n+1)(2n+1)
12
nn-
[(n+ 2)n - nn]. (2k+1) -rr

(2k + 1)7r
313.
fi
k=0
(a1+a2 Ek+ a3 4+
... +a. Er') where ck = cos

.
+/sin 315.
p,, are nth (al +a2 Pi + a3 +onp7-1) where pi, Pz, 1=1 roots of (L. 318. The solution of Problem 223. Adding 1 to all elements of the determinant
a1 0 0 a2
0
0
0 , we get the determinant A.
n n
Aik,
0 ... an
We have A ai a2 ...a n+
n n
Aik=
EE
k=1 i =1
k=1 i =1
E E
( .1 72 ++7 a2. .an 7+ an )
The solution of Problem 250. Denote by A the determinant to be evaluated. We have A = (a,- x) (a 2- x). . . (an - + x A = (a,- y) (a2 - (an Y)+Y
E E
Aik,
A ik
where Aik is the sum of the cofactors of all elements of A. A is readily determined from the system of equations.
7*
196 323.
(a a k)
n (bi_bk)
where f (x)= (x + bi)...(x+ bn). 325. 326. [c +1/ c2 4ab]n + [c V c2 4ab]n c2 -4ab [p + p2 4q]n + [p 1/ p2 4q]n
2n
327. xn + chxn + a3xn + +a., where ak is the sum of all kth-order minors of the determinant an a12 ... a21 a22 an
ant
ant ann
obtained from it by deleting n k rows with indices a2, , a,_k and columns with the same indices.
328. (n + 1)" 1. 329. (xn)n+ 1. 330. (x2-12) (x2-32)...[x2 (2m 1)2] if n= 2m, x (x2-22) (x2 -42)...(x2 -4m2) if n= 2m+1. 331. (x + na n) [x + (n 2) a n+ 1] Ex + (n 4) a n+ 21. .(x na).
n (n-1)
[1! 21. . (n 1)!]2 [(n 1)!]n,

333*n! (n+1)!...(2n-1)! '
332. (-1)
334.
A(a1. a 2,
, an) A (b1, b2, , bn) where A is the Vandermon(1, 2, ..., n)
de determinant.
CHAPTER 3 SYSTEMS OF LINEAR EQUATIONS

335. x1=3, x2 =x3=1. 336. x1=1, x2 =2, x3= 2. 337. x1=2, x2 = 2, x3=3. 338. x1=3, x2 =4, x 3=5. 339. x1= x2= 1, x3=0, x4 =1. 340. x1= 1, x 2 =2, x3= 1, x4= 2. 341. x1= 2, x2 =2, x3=-3, x4=3. 2=2, x3=1, x4= 1. 342. x1=1, x 343. x1=2, x2=x3=x4=0.
344. x1 =x2 =x2=x4= O.
345. x2 =1, x 3= 1, x3=0, x4=2,
197
346. x1 x2 x s x s x 5 O. 347. x1=x2 =x3=x4=0. 348. x1=1, x2 = -1, xs=1, xs= -1, xs=1. 349. x1-x2-x3-x4-x5-0.
350. x1=1, x2 = -1, xs=1, xs= -1, x6=1. 351. x1=0, x2 =2, x5= -2, x4=0, x6 =3. 352. x1=2, x2 =0, x3= -2, x4= -2, x5=1. 353. The determinant of the system is equal to zero, since the system has a nonzero solution. 354. The determinant of the system is equal to -(a2 +/.2-1- c2 +d2)2.
a
E
k=1
ak- aiUn -1) a + (3.1
355. x,356. xi = 357. xi = (t
(cc -13) [(n -1)cx+P]

f (Pi)
f (x)=(x-b1)(x-b2). (x-bn), where <P' (Pi) (P (x)=(x P1) (X -P2). (x-1 3.).
f (t) cq) f , (cci) where f (x)=(x-a 1)-cc5).. .(x- an).
358. xs= E
ui
f,(cci)
cps, i where f (x)=(x- al) . . .(x - an);
t_i
(Ps, i=E 0,0....at n _ s; the sum is taken over all combinations t1, t2, . of 1, 2, ..., 1, i+ 1,. . n.
359. xs=
E
i=i
(Pt, s where f (x)=(x - cci) (x an) (-x ctn);
pi,s=E ati Oita Cln _ ;

here, the sum is taken over all combinations t1, t2, (n-i of 1, 2, ..., s-1, n. s+1, (-1)n an _i 360. xiwhere xn+ai xn-1+ ... +an n1 = (x -1) (x- 2). .. (x - n). k k 361. Cm C n 365. (a) It does not change or it increases by unity; (b) it does not change or it increases by unity or by two. 366. 2. 367. 3. 368. 2. 369. 2. 370. 3. 371. 3. 372. 4. 373. 3. 374. 2. 375. 3. 376. 5. 377. 6. 378. 5. 379. 3. 380. 4. 383. The forms are independent. 384. 23/1-ya -y3=0. 385. ys+3Ya-Ys=0, 2Y3.-Y2-y4=0. 386. The forms are independent.
198
PART
III. ANSWERS AND SOLUTIONS
387. 3,1+1 ,2 y2 y4 =0. 388. Y1 Y2+Y2=0, 5y1-4y2 +y 4 =0. 389. The forms are independent. 390. The forms are independent. 391. y1-FY2 Y3 Y4=0. 392. 2Y1 Y2 y3=0. 393. 3y2 Y2 Ys=0, Yi Y2 y4=0. 394. The forms are independent. 395. yi ya y3 y4 =0 396. 3y1-2Y2 Ya+Y4=0, Yi Y2+ 2y3 y5-0 397. A=10, 3y1+2Y2-5.Y2y4=0. 398. x3=2x2 x1, x4=1. 399. A=5. 400. The system has no solutions. 401. x2 =1, x2 =2, x3= 2. I lx2 x, 402. x2 =1, x2 =2, x3=1. 403. xi = 7 , x2= T 404. The system has no solutions. 5 4 405. x2 =0, x2 =2, x3= x4 = -3 ' 3 ' 406. x1=-8, x2 =3+x4, x3=6+2x4. 407. x1=2, x2 =x2 =x4=1. 408. x1 =x2 =x3=x4 =0. 3x2 -13x, 19x, 20x 4 409. x1 = , x2 17 17 7 5 X5. 410. x1= 6 - x2 x3, x2 = x2 + x2, x4= 411. x1= 16+x2+x4 +5x2, x2 =23-2x3-2x4 -6x6. 4x4 +7x5 4x, + 5x, 4x4 -5x5 412. x 1 = , X2 = , x2 8 8 8 413. x1=x2 -=x2=0, 1+x, 1+ 3x3 +3x, 5x5 414. xi = , XI= 3 3 2+x2 1+ 3x, 3x4 5x, 415. x1 = 3 , x2 6 416. The system has no solutions. 417. The system has no solutions. x, x, x5 418. x1= , x 2 = 1 , x,=0, x4 = 1 6 ' 6 6 420. The system has no solutions. b2+c2 a2 a2 +c2 -62 a2 + b2 c2 421. x= y 2ac 2bc ' z 2ab A+1 1 (A+ 1)2 422. If (A 1) (A+ 2)0 0, x= A+2 , Y= +2 ' z=A4-2 If A=1, the system has solutions dependent on two parameters. If A= 2, the system has no solutions. 419. xi
1+5x4
- 7X4 X2 =
X2 =
1 + 5X4
199
423. If (A 1) (X + 3) 0 0, x v+A-1 A+ 3 Y=
2A + 1 t z= +3 ' If X=1, the system has solutions dependent on three parameters. If A= 3, the system has no solutions. 424. If a, b, c are all distinct,
A2 + 2A+ 2 +3 A3 +3A2 +2A+1 3
x=abc, y=(ab+ac+bc), z=a+b+c.

If among a, b, c two are equal, the solutions depend on one parameter. If a= b= c, the solutions depend on two parameters. 425. If a, b, c are all distinct, (a d)(c d) (a d) (b d) (bd)(cd)
z= y (a b) (c b) (a c) (b c) ' (b a) (c a) , If a= b, a 0 c, d= a or d=c, the solutions depend on one parameter. If b=c, a 0 b, d= a or d=b, the solutions depend on one parameter. if a=c, a= b, d= a or d=b, the solutions depend on one parameter. If a= b = c= d, the solutions depend on two parameters.
In all other cases, the system has no solutions. 2b 1 1 _ 2ab 4b + 1 426. If b (a 1)0 0 , x
b (a-1) ' Y = b ' z b (a-1)
If a=1, b= , the solutions depend on one parameter. 2 In all other cases, the system has no solutions. 427. If b (a 1) (a+2)00, xz
(a 1) (a + 2)
'
y b (a 1) (a + 2)
If a= 2, b= 2, the solutions depend on one parameter. If a=1, b =1, the solutions depend on two parameters. In all other cases the system has no solutions. 428. If (a 1) (a+2)= 0, x=
ma+ m- np
(a+2)(a-1) '
Y = (a-I-2)(a 1) ' z= (a+2) (a-1) "

If a= 2 and m+n+ p = 0, the solutions depend on one parameter. If a=1 and m=n= p, the solutions depend on two parameters. In all other cases the system has no solutions. a2 (6-1) b (a2 1) a -1 429. If a (a b)0 0 , x= Y
na+nmp
pa+pmn
ba
a (a b) ' z a (b a) '
If a= b= 1, the solutions depend on two parameters. In all other cases the system has no solutions. 430. A= A2(X-1). For A=0, A=1, the system is inconsistent. 431. A= 2A. If A00, x=1X, z=0. If A=0, x=1, z=0, y is arbitrary. 432. A= (k 1)2 (k+I). If k = 1, the solution depends on one parameter. If k= 1, the system is inconsistent.
200
433. A =a(b 1) (b + 1). If a=0, b=5, y= 1 , z= 4 , x is arbitrary.
If a=0, b # 1 and b 5, the system is inconsistent. If b=1, z=0, y=1ax, x is arbitrary. If b= I, the system is inconsistent. 434. (a) A= m(m+2). For m=0 and m= 2 the system is inconsistent. (b) A=m(m2-1). if m=0, m=1, the system is inconsistent. If m = 1, the solution depends on one parameter. (c) A= )(A-1) (A+1). If X-=1, A= 1, the system is inconsistent. If A=0 the solution depends on one parameter. 435. (a) 0=3(c+ 1) (c 1)2. If c= 1, the system is inconsistent. If c= I, the solution depends on two parameters. (b) A= (A-1) (A-2) (A-3). If A=2, A=3, the system is inconsistent. If A=1, the solution depends on one parameter. (c) A =d(d-1) (d+ 2). If d= 1, d= 2, the system is inconsistent. If d=0, the solution depends on one parameter. (d) A=(a 02(a+ 1). If a= 1, the system is inconsistent. If a=1, the solution depends on two parameters. x 436.
xl
y 1 yi 1 =0.
1
xo Ya
437. If and only if
xi. h.
Xs
X3
1
1
I
y2
y3
=0.
438. If and only if al b1 ao 122 a3 b3
C2
= 0.
c3
439. If and only if
4+4 xo
x2 +Ya
Yo -Fy!x1 Yi
xs Y2
X3 y3
=0.
4+4
442. y =x2 -1. 443. y
440. (x-1)2+(y 1)2 = 1. 441. y2 y =0. xn )0 -1 ... x 2 x 1
Yo 4
yn .X7i x n-1.
2
xo 1 =0
Xn Xn 1
444. If and only if
x1 Y1 z1 1
X2
x3
Ya Z2 I y3 Z3 1
=0.
xa ya Z4 1
445. xt+ys+exyz=0.
201
446. If and only if the rank of the matrix (Xl yi 1

X2
Y2
Xn Yn 1
is less than three. 447. If and only if the rank of the matrix ( a1 b 1 c 1 a2 b 2 C2
an b n cn
is less than three. 448. In one plane if and only if the rank of the matrix X1 y i Z1 1
X2
(
Y2 Z$ 1
xn Yn zn 1
is less than four; on one straight line if and only if the rank of the matrix is less than three. 449. All planes pass through one point only when the rank of the matrix
1 1 D Al B 1 C
A2 B2 C2 D2
A B C D is less than four; through one straight line only when the rank of this matrix is less than three. 450. n al n-1 a1n all a a21 an a2, n-i
,
= 0.
ant
ant . . . an, n-i ann
453. No. 454. For example, ( 12100 1 2 0 1 0 5 6 0 0 1 455. Yes. 456. Solutien. Let
( 11 C( C1.2 'X in
A11 A1,
Air
A=
rxsu.
06 22
ocan ccrn
B=
A91 An
Aar
an are
art
an
An'
202
A11
as,
E Al, as 5=1
BA
I s =I
Ars asi
(
A s asn s=1
It is immediately obvious that the rows of the matrix BA are solutions of the system. Besides, since I B 1 00, A=B-1(BA), i. e., the solutions written down by the matrix A are linear combinations of the solutions written dcwn by the matrix BA. 457. Solution. Let
a llan a1,, III 112 Yin ,
A=
( am.
c(22 (X2n
C=
Y21 Y22 Y an
ari
ant ccrn
In a -
Ira Yrn Y Y 12 21 -1 ir r
Since C is a fundamental system of solutions, and so on, that is, A=BC, where B An Ala
Art
A A1,
Ars Arr.
On the other hand, A is also a fundamental system of solutions, and therefore I B100. 459. For example, x1=4-1-c2 +5c2, x2 = 2c1-2c2 -6c3, x3=c1, x4 = c2, x2 = c2 (see the answer to Problem 454). 460. xi= 1 1 c, xa= c, xs= 7 c. 461. No. 408. x2 =x2 =x3=x4=0. No. 409. x1=3c2 +13c2, x2 =19c1+20c2, x3=17c1, x4 = 17cs. No. 410. x1=c1+7c2, x2 = 4+ 5c2, x3= c1, x4=2c2, xs=6cs. No. 412. x1=c1+7c2, x2 =c1+5c2, x3=c1-5c2, x4= -2c1; xs=8c2. No. 413. x1=0, x2 =0, x3=0, x4 = c, xs=c. 462. x1= 16 + ci+ c2 + 5ca, x2 =23 24 2c2-6c3, x3=c1; x4 =c2; xs =c3 463. No. 406. x1= 8, x2 =3 + c, x3=6+2c, x4= c. No. 414. x1=c3, x2 =2+4+c2-5c3, x 4 =c2, xs= 1 +3c3. No. 415. x1=1 +2c2, x2 =1 +4c2 +5c2, x3=2c1, x4=2c2, xs=1+6c2.
CH. 4. MATRICES
203
CHAPTER 4 MATRICES
464. (a) ( 3 1) , 5 1 (6 2 1 1 1 (c) 6 8 1 4 (f) (b) (-9 13 \
15 4 i ' 0 0 0 ) 0 0 0 0 0 0 , (e)
, (d)
1 9 15 5 5 9 12 26 32
2+2ac ( a+b+c a' +b2+c2 b a+b+c b2+2ac a2+62+c2 a+b+c a+b+c 3 (7 4 4 ( 15 20) , (c) 465. (a) 9 4 3 , (b)
3 3 4 (d) (0 1 , (e) 20 35 ( 3 \4 cos ncp sin ncp \ cos ncp j sin ncp
2 8/
( 1 466. cc
o n) V 1
= 1+
I cos cp since
k- sin cp cos cp
where tan cp = c` . Hence, n 1 n = ( 1 as( cos ncp sin n cp sinncp cos ncpj lira n cp =a urn a cpo tan y The-
n The limit of the first factor is equal to 1. refore

IX )n
n> co
lim
n
cc n 1
( cos sin a
sin a cos a
467. (a) (A +B)2= A2+ AB+BA+B2= A1+2AB+B2.
(b) (A+B)(AB)=A2 AB+BAB2=A2 B2.

(c) The proof is by induction. ( 10 4 7 468. (a) 6 7 14 4), 5 4 (b) 0
0 0
.
0 0 0 0 0 0
204 469.
(a) t x
2y = (x y)E+ y A , k y x 2y ) x y\ =(xy)E+yA , 0x/ 0 x y 0). u v 3t 3x u t 3y v t
(b) (
(c)
470. (a)
( 5 1 3 80 3 2 1 2
(b)
i00\ \ o 0i
471. It isverified-by direct computation. 472. The polynomials F (x)= ao+ aix+ . . .+ awe' I, such that F (A) =0, exist because the equality F (A)= a0E+ aiA+ . . .+ i 2 ,,,Ain =0 is equivalent to a system of n2homogeneous linear equations in m+ 1 unknowns ao, al, .. ., am, which system probably has nontrivial solutions for ni n2. Let F (x) be some polynomial for which F (A) =0 and let f (x) be a polynomial of lowest degree among the polynomials having this property. Then F (x)= f (x) q (x)+ r (x), where r (x) is a polynomial of degree less than that of f (x). We have r (A)= = F (A) f (A) q (A)= 0 , hence, r (x)= 0, otherwise there would be a contradiction with the choice of f (x). Thus, F (x)= f (x) q (x). 473. Let
,, ( a11 a12 a1 . , B= A= . kn. bnn am. an, arm Then the sum 'of the diagonal elements of the matrix AB is equal to
1=1 k=1 the same. Hence, the sum of the diagonal elements of the matrix ABBA is equal to zero, and the equality ABBA=E is impossible. Remark. The result is not valid for matrices with elements of a field of characteristicp 00. Indeed, in a field of characteristicp, for matrices of order p, / 0 1 0 ... 0 \ / 0 0 1 ... 0\ A= \ 0 0 0 ... I / \ 0 0 ... 0 l we have ABBA= E. 474. (E A) (E+ A+ A2+ ...+ Aki)=E_Ak =E. 475. b (a c a j ' bc= a2 . , / 0 0 0 ... 0 0 ... B= 0 2 0 ...
I
E aikbki The sum of the diagonal elements of the matrix BA is exactly
0\ LO \ 0
\0 0 0 ... p 1 0 i
CH. 4. MATRICES
205
476. If A2 =0, then A2 =0 also. Indeed, if A2=0, then 1 A 1=0. Hence (see 471), 'A2 = (a+ d)A, 0= A3= (a+ d)A2=(a+ d)' A, whence +d=0 and
A2 = 0.
477. E, ( a
b) c a
a2=1bc .
478. If A=0, then Xis any matrix. If 1 A 100, then X=0. Finally, if Al =0, but A 0 0, then the rows of matrix A are proportional. Let a : (3 be the ratio of corresponding elements of the first and second rows of the matrix A. Then tiflx X= for arbitrary x, y. k 133' ay 479. Let A=( c d ) (1) If A00, but a+ d=0, adbc=0, then there are no solutions; (2) if a+d00, (a d)2+4bc=0, (ad), b, c are not equal to zero simultaneously, then there are two solutions:
ab
2b\ 3a+d 1 X= , 2c a+3d 2 V 2(a+ d) (3) if a+ d00, adbc=0, then there are two solutions: ab\ 1 X= Va+d c dl' (4) if adbc0 0, (a d)2+ 4bc0 0, then there are four solutions: A2+ad 1 ( 2 X2 a+d b
2 where 2= + 1 I a+ d+ 2 1/ ad bc ; (5) if a d=b= c=0, then there are infinitely many solutions: X= + - V a.E and X=
( X z
x_ ( 2
1 0 0
, where x, y, z are connected by the relation x2+yz= a. 2 (b) 1
480. (a)
1)'
2 1 0 7 2 1
ad be
/1 0 0 \0
d b a ' c
3 1 0 0 71
-
11 2 1 0 1
1 1
38 \ 7 2 / 1 1
1 11 \
(c)
(d)
(e) (
1 1 1
4 5 6
3 3 4
(f)
\il
1 1
1 1
1 / 1/
206 ,
2 3
1 2 19 14 2n
1
0 0 3 2
0\ 0 4 / 3
(g)
( 31
23
1
2n
(h)
1 n1
1 1
2n ...
1 \
1
1 1
1
4
1 1 1
... 2nl
1
e-2/1+2
1 (i)
2 e n-1-1 ,-2n+2 ... ,(n -1)2
(j)
1 n+1
1 (n 1) 1 (n-2) ... 1 1 / 1n I 1. (n 1) 2. (n 1) 2. (n-2) ... 2.1 \ 1 (n-2) 2. (n-2) 3. (n 2) ... 3 1

21 \ 11 2 2 2n2 2+n n2 2n2 2+ 2
2+n2 2 2
3. 1 2 2
2 n2 )
nI/
(k)
1
2n3
b2 cn Nen+ d --cnI 1 bi b2 .. hn where d= a bici b 2c a .. . bnen,

bi en
\
1
(1)
1 d
hi c i - -F. d
b2 ci
6,, ci
c1 \
c2 1
b1 c2 b2Ca d
bn C2
f _ fo x n x f_ fi x n . . . xn 1 f _ fn_1 xn x n 2 f_ fn ix ni .. tt 1 . _ foxn 1 f_fi, x

...
(
xn
xn i x
(m) f
fn-1
where fo= ao, fi= aox+
, .f_ aoxn 1+ ...+an_i, f= ao x" + aixn +

AI
2X1 _ 1( A
...+ an,
Al 0 . . . 0 (n) 0 as 0
AlAa Al an
a2 . . . a2 An
. . .
an
analAnA2 an
CH. 4. MATRICES
207
tk=1 +Xi+Az + - . +An ; (13-1+XB-' UV B-1 -AB-1U ) 1 , A(o) a- VET-1 U -A vB-i A 481. (a) (20 /1 1 0 (c) 0 \0 (d) / 24 -I 1 1 0 0 -2 3 8 )' (b) 0 -3 2 - 4 5 -2 0 -5 3 0 0 0
\
0 ... 0 0 0 -1 -1 -1 ... 0 0 0 1 -1 ... 0 0 0 0 0 13 -18 0 ... 0 1 1 0 ... 0 0 1 (e) X= E -
-1 2/
(
n-1
\ 34
(f)
1 1 1
1 1 1 1
b , (g) X does not exist. 1+a -2a 1-26 482. It is sufficient to multiply the equation AB=BA on the right and on the left by A -1. 483. 484. If A3=-E, then I A 13 =1, and, by virtue of its real nature, I A 1=1. Then, equating A -1and A2, we readily find that A=E or (c a b d). a+ d= -1, adbc=1. (a b) , a2+bc= +1. 485. A= + E or A= c-a Set A= 1 0 1 \ ( a b = a E+ b I, where /= k 1 o \ -6 a Then P= - E and, hence, the correspondence aE+ bI-> a+ bi is an isomorphism. 486. a+bi c+di) = aE+ + dK , -c+ di a-bi 0\ / 0 1 0 i \ i where (i 0 f. Then 7=(0 -4/ ' 41.= \ -1 0), K= p=J2=K2= -E, IJ= - JI=K, JK= - KJ=I, Kl= -IK=J. Whence it follows that the product of two matrices of the form a+ + T + dK is a matrix of the same type. The same holds true for a sum and a difference so that the 487. Set
208
a+bi c+di c+ di a bi =a2+ b2+ c2+c/20 0 as soon as A00. Hence, every nonzero matrix has an inverse, and from the equality A1A2=0 (or A2A2=0) for A100, it follows that A2=0. The ring of matrices under consideration is a realization of what is called the algebra of quaternions. 488. (aiE+ b1I+ c11+diK) (a2E+ b21+ c21+d210=(aiaa bib a cica- 442)E+(aib2+ bia2+ 4(4 dic2)1+(aic2bid2+ cia2+02)J+(aid2+ + bic2 cibg d ia2)K. Taking determinants, we get (a1+191+ cl+ 0 (a: + b:+ + 4 =(ala2 bib 2 cac2 did2)2+(aib2+ bia2+ cid2 dica)2 + (alc2 bid2+ c1a2+ 02)2+ (aid2 +bica c1b2+ d,a2)2. 489. Interchanging two rows of a matrix is accomplished by premultiplication of the matrix set of matrices at hand is a ring. Furthermore, (A 1=
1 ..
Operation b is accomplished by premultiplication of the matrix
i1
1 ... a or 1
a ...
1/
CH. 4. MATRICES
209
Operation c is accomplished by premultiplication of the matrix /1
1 a 1
The operations a, b, c on columns !are accomplished by postmultiplication of the same matrices. 490. Any matrix A can be reduced to diagonal form R by elementary transformations a, b, c on the rows and columns. Therefore,' for the given Urn AV1V2... Vk, matrix A there is a matrix of the form R such that R= Vk are matrices of elementary transformations. where U1, ..., U., V1, They are all nonsingular and have inverses. Consequently, A=PRQ, where P and Q are nonsingular matrices. 491. By virtue of the results of Problems 489, 490, it will suffice to prove the theorem for diagonal matrices and for matrices corresponding to the operation a, because matrices corresponding to the operation b have the required form. It is easy to see that the operation a reduces to the operations b and c. Indeed, to interchange two rows, we can add the first to the second, and then from the first subtract the second, then add the first to the second and, finally, multiply the first by 1. This is equivalent to the matrix identity Eeii ekk+ +eik+eki=(E-2ekk) (E+eik) (Eeki) (E+eik). The theorem is, obvious for diagonal matrices: a1e11+a2e22+ +anen. = (E+ (02 1) en) (E+ 1) e22) (E+ (a. 1) en,) 492. Let A=P1R1Q1, B=P2R2Q2, where Pi, Ql, P2, Q2 are nonsingular matrices and R1and R2 are matrices having, respectively, r1and r2 units on the principal diagonal and zero elsewhere. Then AB=P1R1Q1P2R2Q2and the rank of AB is equal to the rank of RICR2, where C= Q2P2is a nonsingular matrix. The matrix R1CR2is obtained from the matrix C by replacing all elements of the last rtr1rows and n r2columns by zeros. Since striking out one row or one column reduces the rank of a matrix by no more than unity, the rank of R1CR2 is not less than n(nr1)(nr2) =r1+r2 n. 493. It follows directly from the proportionality of all rows of a matrix of rank 1. 494. On the basis of the results of Problem 492, the rank of the desired matrix A is equal to 1 or 0. Hence
Al V1 Al [42 A2 [1.8 A3 112 Al [-L3
12 1-13 )
A=( A2
A3 ill
A3
(1 3
Direct multiplication yields 0 = A2= (A1P-1 + X2!12 + 34 DA whence it follows that A1[1.1+ 12[12+A3113=0
210
495. Let A yield a solution to the problem different from the trivial A= E. Then one of the matrices A E or A+ E has rank 1. Let
Al t Al (as Al !la A+ E=( A2 [Ll A2 tL2 A2 1 3 ) =B. 1
A3 1-41 A3 1/2 A3 1 23
Then A2 = E 2B + B2= E + (Aik-ti+ 2[1, 2+ ASP. 3 2) B, whence it follows that for A2=E it is necessary and sufficient that the condition A11 12+ A31-1. 3= 2 1-1+ X21 be fulfilled. The second case is considered in similar fashion. 496. Let there be adjoined to the matrix (A, B), where ll aik
A= arn,. amk Ca , B b mi b ms
\
the column C=
cm l
, the adjoining of which to the matrix B does not
increase its rank. Then the system of linear equations biiYi+ +bis Y s= al. + + bmsYs = Cm is consistent, but then also consistent is the system
nazi+ . . + aikzk + bnYi+ +bisYs=ci,

amt x1+ -Famk XkbmiYi+ +bmsYs= Cm.
Consequently, the rank of the matrix (A, B) is equal to the rank of the matrix (A, B, C). Now suppose that the columns of the matrix B are adjoined to the matrix A gradually, one at a time. The rank can then increase by unity by virtue of what has just been proved only when the rank of B increases. Hence, the rank of (A, B) the rank of A + the rank of B. 497. Let the rank of (E+A)=r1, the rank of (E A)= r2. Since (E+ A)+ +(E A)=2E, r1+r2..n. On the other hand, (E+ A) (E A)= 0, therefore 0 ?:ri+r2 n. Hence, r1d-r8=n. 498. The rank of the matrix (E+ A, E A) is equal to n. From this matrix choose a nonsingular square matrix P of order n and let its first r columns belong to the matrix E+A, and let the other nr columns belong to E A. Then, by virtue of the fact that (E+ A) (E A) =0 we have
(E+ A) P (Int qm (E A) P (
0 0 qn,r-Fi qnn
0 )
0 0
0 0 91,r+1 gin
CH. 4. MATRICES
2H
Combining these equations, we get 2P = ( 911 9ir 91, r+i Stn
gni qnr qn, r +1 qnn
Subtracting them, we obtain 2A P =

911 qv'
( qni r +1 q1n
r
gm- qn,
9nn
= 2P
1/
Whence follows immediately what we sought to prove. 499. If AA -1=E and both matrices are integral, then 1 A IxIA '1=1, whence it follows that I A I = 1 because I A I and I A 1 -1are integers. The condition I Al= +1 is obviously also sufficient for the matrix A -1to be integral. 500. Let A be an integral nonsingular matrix. There will be nonzero elements in its first column. By multiplying certain rows of the matrix A by I, we can make all elements of the first column nonnegative. Choose the smallest positive one and subtract its corresponding row from some other row containing a positive element in the first column. We again get a matrix with nonnegative elements in the first column, but one of them will be less than in the original matrix. Continue the process as long as possible. In a finite number of steps, we arrive at a matrix in which all elements of the first column, except a positive one, are zero. Then, by interchanging two rows, carry the nonzero element of the first column into the first row. Next, leaving the first row fixed, use the same operations to obtain a positive element on the diagonal in the second column, all elements below it being zero. Next turn to the third column, etc. The matrix will finally become triangular. Then, by adding each row an appropriate number of times to the above-lying rows (or subtracting each row from them), we reach a situation in which the elements above the principal diagonal satisfy our requirements. The foregoing operations are equivalent to premultiplication by certain unimodular matrices, whence immediately follows the desired result. 501. Let A =PiRi=P2R2, where the matrices P1R1and P2R2 satisfy the requirements of Problem 500. Then from the equation P21P1-=R2RT1 it follows that the integral unimodular matrix C=Pil P1is also of triangular form.
212 Let
/ an a12 ain a22 a2n

. . /
b11 b12 ... bin
b2,
R2 =
b2,,
R1=
ann
C12 Cl,, C22 C2n
b .an
C=
. . . .
can
Then from the equation /22 = CR1we first of all conclude that b11=c11a11, ...,b,,.= ca., whence it follows that all CHare positive. But CnC22 Cnn = = 1 a 1= 1, hence cu.= c22= ...=c,,=1 and aii=bli. bit--12 Furthermore, b 1 2 = -11-12 a - 1--12a a12 +C 22 = -12-22, whence C12 , 22 But 0 -<..b12 < b22=a22, 0 --<-.a1.2< a22, hence, 1 C12 1 1, and therefore C12=0. Thus, comparing successively (by columns) the other elements in the matrix equation CR1=R2, we come to the conclusion that all cik =0 for k> i, that is, C=E. Consequently, RI= R21 Pi=P2. Thus, in each class there will be one and only one matrix of the form R. The number of matrices Rwith given diagonal elements d1, d2, d .o is evidently equal to d24...dn n -1and so the number of matrices R with a given determinant k is equal to F,,(k)=E d2d:...d."-1, where the summation sign E is extended over all positive integers d1, d2, d,, satisfying the condition did2...d.=k. If k= ab, (a, b)=1, then each factor diin the equation k= =d2d2...d,, is uniquely factored into two factors oci, so that (xioc2...ocn=a, 131132 -13n= b. Hence, Fn (k)=
d,.dz do =k
d2 ... 4-1 Pa P.:
I
an -= a P. (32 . 13n= b
.22:
14-1
E
czi ac ... Gen = a
cx2
1
PI
E
P2
p2
Pn=b
[4-1= F.(a) Fn(b).
From this we conclude that if k=pr' psi is a canonical factorization of k into simple factors, then F,,(k)=F.(pr') It remains to compute Fn(p"1). To do this, break up the sum for the computation of Fn(p"') into two parts, in the first of which dn=1 and in the second of which dois divisible by p, dn=pd'.. This yields the formula F.(pn=
CH. 4. MATRICES =F_1(Pm)+Pn-IF (Pm matical induction, that
213
from which we readily establish, via mathe(pm+n -i_i)
F,,(p")=
(1_0(prrl+2_ tn-11)
(P -1) (P2 -1)
(P"-1 -
502. Choose the smallest (in absolute value) nonzero element of the matrix and carry it into the upper left corner by interchanging rows and columns. Then add the first row and the first column to all other rows and columns or subtract them as many times as is needed for all elements of the first row and the first column to be less than the corner element in absolute value. Then repeat the process. It will terminate after a finite number of steps because after each step the element which arrives in the upper left-hand corner is less in absolute value than the preceding one was. However, the process can only terminate in the fact that all elements of the first row and the first column (except the corner one) will become 0. In the same fashion, transform the matrix formed by the 2nd... nth rows and columns. The matrix will finally be reduced to diagonal form. By virtue of the result of Problem 489, all the above-described transformations are equivalent to postmultiplication and premultiplication of unimodular matrices. 503. Premultiplication of the matrix A'n is equivalent to adding the second row multiplied by m to the first. Premultiplication of BI" is equivalent to adding the first row multiplied by m to the second row.
ab) be a given integral matrix with determinant 1. Divide a c d by c: a= mc+ al, 0' al< I c I, then divide c by ch: c= inlaid- c2, 0 .. c2 < at,
Let U= ( ( al b1 etc., until the division is exact. Then A-m U= Ul= c d ), B-mUi=
= U2= 1/ al
\Cry
b1 )
d2
, etc. We finally arrive at the matrix Uk+ 1of the form

( 0 bk +1 ) .
(a k bk or 0 dk+i)
Ck dk
Then, by virtue of the positivity of all ak,
ck and the unimodularity of Uk+i, we have ak = dk+ 1=1 in the first case, and
ck= -bk+1=1 in the second. Thus, Uk+i=(
1 bk )
0 1
= A k in the first case,
b.
d 0 - 1 ) =A -1B A k -1in the second. The proof of the theand Uk-Fi=( , t dk orem is complete. 504. A matrix with determinant -1 is transformed into a matrix with determinant 1 by multiplication by C. Each such matrix is a product of the powers of A and B. But B= CAC. 505. Let I A 1=1, A2= E, A E. Then (Problem 498) (1 A =P -1 -1 ) P-1
214
PART III. ANSWElS AND SOLUTIONS
for some nonsingular matrix P. Define matrix P so that it is integral with the / 2 smallest possible determinant. Since A+ E=P 0 P ', the matrix A+ 0 +E is of rank 1 and, hence, Al (Li Al tta Ail-L3 A + E= / A2 11 2 A2 I-13 1. A2 11
% A3 41 A3 tit
AS 1L3
Here, XiiLi+ A2(12+ A3(i3= 2 (Problem 495). Since the matrix A+ E is integral, the numbers Ai, A2, Asand the numbers 41, 1-12, 113 may be taken to be integers. Forming a system of equations for the components of the matrix P, it is easy to verify that for P we can take the matrix 0 a , Al / A2 1 u( 11 P 11 a v
where 8 is the greatest common divisor of p.2, p.3 and u, v are integers such that ut.t2 + vp.3 = a. The determinant of the matrix P is equal to 2. On the basis of the result of Problem 500, P= QR, where Q is unimodular and R is one of the seven possible triangular matrices of determinant 2. Consequently, Q 1A Q is equal to one of the seven matrices
1 I )R-1.
1 Of these matrices, there are only three distinct ones, and two of them pass into one another by a transformation via the unimodular matrix. That leaves the two indicated in the hypothesis of the problem. 2 4 6 9 3 506. (a) ( 10 3 , (b) (10 ) , (c) 1 2 3 , (d) 13. 8 3 6 9 507. 45. 508. As a result we get Euler's identity: (al+ b1+ 4) (4+61+ 4)=(aa,+ bb1+ cci)2 + (a1 b2 a2 b1)2 (a 1c2 a2 Cl)2 (b1 c2 b2 c1)2. 509. The minor made up of the elements of rows with indices i1, i2, ..., im and of columns with indices k1, k2, ..., kmis the determinant of the product of the matrix made up of rows 13, i2, imof the first factor by a matrix composed of the columns k1, k2, ..., kmof the second factor. It is therefore equal to the sum of all possible minors of mth order made up of the rows of the first matrix with indices i1, i2, im multiplied by minors made up of the columns of the second matrix with indices k1, k2, .1 km. 510. The diagonal minor of the matrix AA is equal to the sum of the squares of all minors of the matrix A of the same order made up of the elements of the columns having the same indices as the columns of the matrix AA, which columns contain the given minor. It is therefore nonnegative. 511. If all principal minors of order k of the matrix AA are 0, then, by virtue of the result of Problem 510, all minors of order k of the matrix A are equal to 0. Hence, the rank of the matrix A, and also the rank of the matrix AA, is less than k.
R(
CH. 4. MATRICES
215
512. The sum of all diagonal minors of order k of the matrix AA is equal to the sum of the squares of all minors of order k of the matrix A. Also equal to this number is the sum of all diagonal minors of order k of the matrix AA. 513. It is obtained by applying the theorem on the determinant of a proa, a2 an ) duct into duct of two matrices to the product of the matrix
b, " 2 . . bn
its transpose. 514. It is obtained by applying the theorem on the determinant of a product to
a', b',
I a, a2 ... a\( 4 b, 2
b, b 2 . . bnl
- -
an b' n 515. It follows directly from the identity of Problem 513. The equal aia, ... a, is less than sign is only possible if the rank of the matrix b1b.2. 1'2 two, that is, if the numbers a1, a2, , an and bi, b2, ..., bn are proportional. 516. It follows directly from the identity of Probl em 514. The equal sign is only possible if the numbers a1, a2, ..., a and b1, b2, ..., b,, are proportional. 517. Let matrix B have m columns, and matrix C, k columns. By the Laplace theorem, I A I = E BiCi, where Biare all possible determinants of order m constructed from the matrix B, and C2are their cofactors, which are equal (to within sign) to the determinants of order k constructed from the matrix C. By virtue of the Bunyakovsky inequality (Problem 515), I A 18.,E B? E C. 2 But, E BF = I BB I, E Ci =1 CC I 518. Let
( )
( c11 cik ) ( bil. bvn ) , C ; bni bnm Cn1 Cnk
A =(B, C).
The inequality being proved is trivial if m+ k> n; for the case m+k=n, it is established in Problem 517. There remains the case m+ k<n. First assume
that
Eb11 c12
i=
0 for arbitrary j, s. Then
AA=
_C4.8
0
and, consequently, I
CC
AA I= IBBI I CC I.
In the general case, it suffices to solve the problem on the assumption that the rank of the matrix A is equal to m +k, for otherwise the inequality is trivial. Complete the construction of the matrix A to the square matrix (A, D) so that the rank of the matrix D is equal to n mk, and the sums of the products of the elements of any column of D by elements of any column of A are 0. For example, this can be done as follows. First complete construction of A to the nonsingular square matrix e' = (A, D'), which is evidently possible, and then replace all elements of the matrix D' by their cofactors in 1 e' I. The rank. Of the thus constructed matrix D will be equal to the number
216
of its columns n m k, for it is a part of the matrix made up of the cofactors of the matrix e", which differs from the nonsingular matrix (t') -'1in the sole factor I e' I. Denote (A, D) by P, (C, D) by Q. Then, by virtue of the result of Problem 517, I PP 11 1 1 QQ 1. But I PPI= IAAII DD1 and I 0Q1=1 dc" I I DD I. Whence it follows that since I DD 1> 0 ,
I AA I
BB1. ICC 1 .
519. This follows directly from the result of Problem 518 as applied to the matrix A. 520. The determinant of A* A is the sum of the squares of the moduli of all minors of order m of the matrix A, where m is the number of columns of A. 521. This solution is similar to the solution of Problems 517, 518. For a square matrix, the question is resolved by applying the Laplace theorem and the Bunyakovsky inequality. It is advisable to complete the rectangular matrix to a square matrix so that the sum of the products of the elements of any column of the matrix A by the conjugates of the elements of any column of the complementary matrix is equal to 0. 522. Applying the result of Problem 521 several times to the matrix A and taking, for B, a matrix consisting of one column, we get
n
I A*A 1=11 A112 r.-CE I aiiI2 E

i =1 i =1
n
E ain
i =1
whence it follows that

11 A
M".
523. Complete the given determinant A to a determinant Alof order n+ I by adjoining on the left a column all elements of which are equal to _, and a 2 row of zeros on top. Then A= 2 _ 1. Subtract the first column from all coA lumns of the determinant Al. We get a determinant, all elements of which do M not exceed Using the result of Problem 522 yields what we set out to 2 prove.
n
524. The bound n 2 Mn is attained, for example, for the modulus of the determinant 1 1 1 e
1
En-1
... 1 en -1
c(n -
where e = cos
27c + .
sin
22n
525. Construct a matrix of order n=2'n as follows. First construct the matrix ( 1 1 1 . Then replace each element equal tq 1 by the matrix
CH. 4. MATRICES
217 )
and each element equal to I by the matrix 1 (1 1 1 . We obtain a fourth-order matrix,
11
,1
1 1 1 1 1 1 1
1 1
Operating on this matrix in the same way, we obtain an eighth-order matrix, etc. It is readily seen that for the matrices thus constructed the sums of the products of the corresponding elements of two distinct columns are zero. Consequently,
AA=
n 0 ... 0 0 n . , . 0),
(
I AA J = n4,
A iHn n
00
n
n
The following equation holds for the matrix MA:
II MA II=Mn 2
526. We prove that all the elements of a matrix, the absolute value of the determinant of which has a maximal value, are equal to 1. Indeed, if 1 < < ask <1, A>0 and Ask > 0, then the determinant will be increased by substiand A sk<0 , the determinant will increase due to tuting 1 for ask, but if A 1 replacing ask. If A <0, then the absolute value of the determinant will increase when aikis replaced by unity with sign opposite that of Ask. Finally, if Aik =0, then the absolute value of the determinant will not change upon aik being replaced by 1 or 1. We can say, without loss of generality, that all the elements of the first row and the first column of a maximal determinant are equal to I ; this can be achieved by multiplying the rows and columns by 1. Now subtract the first row of the maximal determinant from all other rows. The determinant then reduces to one of order n-1, all elements of which', are 0 or 2. This latter determinant is equal to 2n-1N, where Nis some integer. 527. 4 for n=3, 48 for n=5. 528. For a singular matrix A the result is trivial. Let A be a nonsingular matrix and let A be its transpose, A its determinant and A' its adjoint. 1 / +1 1 ; this follows directly Then A' =AC/1-1C,! where C= from the rule for constructing an inverse-matrix. Therefore, I A' I =An-1 =An-1 A-1A =An-2 A. and (AT =An-1. C(4)-1 C. 529. Let the minor of the matrix A', which is the adjoint of the nonsingular matrix A, be made up of rows with indices is< is < < imand columns with indices Ics <lc, < Let ini.fs.<4,4. 2 < < inbe indices of the rows
218
not in the minor, let k,1< ...<Icnbe indices of columns not in the minor. Multiply the minor at hand by the determinant A of the matrix A:
Ai, k, Ai, k m I
A.
Aim k, Aim kin
= ( -1)
Ai, k,
Ai,k m . . Aim k, Ai m + ,k,
+k m A
Ai, k, Ai,k m
. Ain,k, Aimk,
. . Aik,
m}, k m . A in, k m Ai . . . Ai n km 1
ai, k, aim k, ain k,
ai, km aim kn, ain km
ai, k n
aim kn ain kn
A . . aim +1 k m +, ai m +1 k n ain k,
ain k m+ , ain k aim+ , km+,
ain,+ , k n
_Am ain km+ ain kn
whence follows what we sought to prove. 530, 531. This follows directly from the theorem on the determinant of a product of two rectangular matrices. 532. It is necessary to establish alphabetical ordering of the combinations, i.e., consider the combination 11< i2 < < in, as preceding the combination 11<12< <jn, if the first nonzero difference in the sequence it j1, 12 ..., is negath e. Then each minor of the triangular matrix, the in12 dices of the columns of which form a combination preceding the combination of the indices of the rows, is equal to 0. 533. By virtue of the results of Problems 531, 491 it suffices to prove the theorem for tr;angular matrices. By virtue of the result of Problem 532, we have for the triangular matrix A, ai, ai, m= A aim i
CH. 4. MATRICES
219
534. Properties (a) and (b) follow directly from the definition. To establish Property (c), it is convenient to denote the elements of the Kronecker product, using for suffixes not the indices of the pairs but the pairs themselves. Let
C= (A' A") x (B' B"), A' x B' = G, A" x = H.
Then
Cit lc,, i2 k,
=E
i =1
EGh k bk E
,
in
kz
kk a;'4 aii b'
11k = E
1,k
gi, k ,, ik hik, iz kz
k =1
i,k
whence C=G H, which completes the proof. 535. The determinant of the matrix A X B does not depend on the way the pairs are numbered, because a change in the numbering results in identical interchanges of rows and columns. Furthermore,
A x B= (A x Em) (E,,x B).
Given an appropriate numbering of the pairs, the matrix A x Em is of

,A I A
the form
, the matrix A being repeated m times. Conse-
A' quently, the determinant of A xEm is equal to I A Im . In the same way (but with the pairs numbered differently), we see that the determinant of En x B is equal to 1./31n. Hence, 1 A x B1=1A 1' I B1'. 536. An element of a row with index a and of a column with index p of the matrix Cik is
Inn
c(i_i)m+a,(kOmi-p=
m s =1
s bs, (k 1) +
a(i _1) m+ (j in v b(j 3.)m+v,(k 0 n2+ (3. v=1 But the inner sum in the last expression is an element of the matrix Aii Bik taken from a row with index a and a column with index B. Thus;
,
Cik =
ij Bjk
j =1 537. For n=1 the theorem is trivial. Assume that the theorem is proved for matrices of "order" n-1 and prove it for matrices of "order" n. First consider the case when Anis a nonsingular matrix:
A11
Al2 A22
Ain A2n
).
C(
A21
Ant
Ana
Ann
220
Multiply matrix C on the right by matrix D, where

/ ED=
42
Ain
E Then C' =CD will be of the form

All C'= A,1 Aryl
0 A'22
...0 AZn
A, '22 Ann
where A;k =Aik-A0A -11 All submatrices of C, D and C' commute with one another. It is easy to see that when this condition is fulfilled the theorem on the determinant of a product of two matrices is also true for formal determinants. The matrix D has a formal determinant E, the actual determinant of D is equal to 1.
de22 41
Hence, I CI=1 C' 1=1 An 1

Ana .
A;,,,
and for the formal determinant of B we will have B= An. B', where B' is the formal determinant of the matrix
By the induction hypothesis,

11;2 . Asn
1B' ...
Ann
and, consequently, I B1=1A111 I B' 1= I C I, which completes the proof. In order to get rid of the restriction I A11 100, the following can be done. Introduce the matrix
C (A)=
All -FA Em An . . . 2 111 A_,, A,, . . A21
A11
Ana
Ann
and denote its formal determinant by B (A). Since I An + XE, I= Vn+ .. .00, I C = I B (A) I . Both these determinants are polynomials in A. Comparing their constant terms, we obtain I CI=1 BI. This completes the proof.
221
CHAPTER 5 POLYNOMIALS AND RATIONAL FUNCTIONS OF ONE VARIABLE

538. (a) 2x2 -7x5 +6x4-3x3 x2-2x+1,
(b) x3 x4-4x3+3x+1.
539. (a) The quotient is 2x2 +3x+11, the remainder, 25x-5.
(b) The quotient is
3x-7 26x-2 , the remainder, 9 9
540. p=q2-1, m=q. 541. (1) q=p-1, m=0; (2) q=1, m= -1/2p. (x 1)(x 2) .. . (x n) 542. (-1 1 2 3 . . . n 543. (a) (x-1) (x8 x2+3x-3)+5,
(b) (x+3) (2x4-6x3+13x2 -39x+109)-327, (c) (x+1+i) [4x2 (3+41)x + ( 1+7 i)]+8-6i, (d) (x 1+2i) [x2 -21x 5-2i] 9 + 8i. 544. (a) 136, (b) 1-44i. 545. (a) (x+ 1)4-2(x+1)3-3(x+02 +4(x+1)+1, (b) (x 1)5+ 5(x 1) 4+ 10(x - 03+10(x-1)2+5(x-1)+1, (x- 2)4- 18(x-2)+38, (d) (x + 2i(x + (1 +i) (x + 02 5(x+i)+7+5i, (e) (x + 1 204 + 1 208+ 2(x 4- 1 2i) + 1. 6 11 1 7 546. (a) + + 24 (X 2)2 (X 2)3 (x ) (x 2)5 1 4 4 2 (b) x+1 (x+1)2 (x+1)3 (x+1)5 547. (a) x4+11x3+45x2 +81x+55, (b) x4 4x3+ 6x2 + 2x+ 8.
548. (a) f (2)=18, f (2)=48, f" (2)=124,
f - (2)=216, fiv z)= 240,

(
f v(2)=120; (b) f (1 +2i)= 12 2i, f' (1 +21)= 16 +81, f" (1 +2i)= 8+ +30i, f" (1 + 2i)= 24 +301, f TV(1 + 2i) = 24. 549. (a) 3, (b) 4. 550. a= 5. 551. A = 3, B= 4 . 552. A = n, B = (n+ 1).
222
555. For f (x) to be divisible by (x-1)k +2, it is necessary and sufficient hat f (1) = no + al + + a = 0 and f' (x) be divisible by (x-1)k ; for this it is in turn necessary and sufficient, given the condition f (1)=0, that fi(x)--nf (x)-xf ' (x) be divisible by (x- 1)k. Regarding f1 (x) formally as a polynomial of degree n, we repeat the same reasoning k times. 556. a is a root of multiplicity k + 3, where k is the multiplicity of a as a root of f ' (x). 557. 312562 +108(25 =0, a00. 558. b=9a2, 1728a' + c2 = O. 559. The derivative xn-m [nxm + (n- m)a] does not have multiple roots other than 0. 560. Setting the greatest common divisor of m and n equal to d, m=dm1, n= dni, we get the condition in the form ?T. ( - l)nx (ni -nti)n= -11h nr ini an= =
562. A nonzero root of multiplicity k - 1 of the polynomial
at xlh+ a2xP.+ . . . + ak xPk
satisfies the equations

xPi + a2 xP-1- . . . + ak x Pk =0,
Pt at xPl+p2 a2 xlh+
+pk ak x P k =0, +p2 k ak x Pk =0, pk -20k x P k =13
P1 atxPl +14 a2xPz +

pk-2
a1
+pk 2 02 .x ri .
2
whence it follows that the numbers ai x=, a2 x1", ak xPk are proportional to the cofactors of the elements of the last row of the Vandermonde determinant 1 A = Pi
pif -1
1
Ps
1
Ps
...
Pk Pl ir l
pr iP1
so i
It is easy to verify that A Di = 11 (pips)=V (Pi). From this it follows that the numbers ai xPiare inversely proportional to (P' (Pi), i.e. a1xP' (PO = a2 x Pa p (PO= . . = ak x P kcP' (Pk). All the foregoing reasoning is invertible. 563. If f (x) is divisible by f' (x), then the quotient is a polynomial of degree 1 one with leading coefficient n ,where n is the degree of f (x). Therefore, nf (x)=(x-xo) f' (x). Differentiating, we get (n- 1) f' (x)= (x- xo) f" (x), and so on, whence
f (x)=
(x n!xo)n f (x)= a(x-
The converse is obvious.

x
223
xn
+ . . . + must 564. A multiple root of the polynomial f (x)= I + 1 n1 also be a root of its derivative
f' (x)= 1 + T +...+ (n
xn 1
f (x) nl
xn
Hence, if f (x0)= f' (x0)=0, then xo =0, but 0 is not a root of f (x). 565. If f (x)= (x x0) 4 (x), where f1 (x) is a fractional rational function which does not vanish for x=x0, then direct differentiation yields .(x o)_ =f (k-1) (x0)=0, f (k) o) O. f (x0) =1 Conversely, if f (x 0) = f' (x0) = = f (k (xo)=0 and f (k) (x0)0 0, then f (x)= = (x x0) kJ: (x), f1(x0)00because if it were true that f (x)= (x xo)m q (x), q (x0) 00 for m k, then the sequence of successive derivatives vanishing for x would be shorter or longer. 566. The function
g (x)=
t.1) (x)
W
(x)
=f (x) f (xo)
f' (xo) (x x0) . . .
f (n) (x)
n1
(x x or
satisfies the condition
g (x o)= g' (x 0) = . . . = g(n) (x 0) = O.

Consequently, 4,(x)=(x xo)n+ 1F (x), where F (x) is a polynomial. This completes the proof. 567. If A. (x) fa (x0)f2 (x) A (x.) is not identically zero, then we can take it that 1 1(xo) 00. Consider the fractional rational function A (x) A (xo) It is not identically zero and has xoas a root. The multiplicity of this root is higher by unity than the multiplicity of xoas a root of the derivative equal to whence immediately follows the truth of the asserM. (x)P tion being proved. 568. Let xobe a root of multiplicity k for If (x)]2 f (x) f' (x). Then f o) 0 0 because otherwise xowould be a common root off (x) and f' (x). From the preceding problem, xo will be a root of multiplicity k +1 of the polynomial f (x) f' (x0) f (x0) f' (x), the degree of which does not exceed n. Hence, k+ 1
fi (x) f (x) (x) f '1(x)
f (9 fa (x o)
n,
569. The polynomial f (x) f' (x0) f (x0) f' (x) must have xo as a root of multiplicity n, that is, it must be equal to A (x xo)", where A is a constant. An expansion in powers of xxo, after the substitution xx0=z, yields (a0 + z + a2 + . . . +an Mai (al+ 2aa z +3a3 z2 + .20= Azn + nan
and ao= f (x0) Whence a2 =

?
().
ay:
2a, , a2
a 31 ' ' an =a/J-1n!
Substituting =a, we get ao
f (x)= [1 +
(x x0) 1
0(2 (x 0)2 + 1 2x
(xx0r1
n!
224
570. For example, S= 21 (20 is not suitable) . 571. For example, = 25

5 (
24
not suitable) . o
/
572. F r example, M=6. 573. For example,

33
(a) x =pi, 0 < p <
(b) x=p, 0<p<1/ 3.
574. For example, (a) x =1 p, 0 <p < ; 4 1)7r +i sin (b) x= 1 + p (cos (2k (c) x=1 + pi, p< 575. Expansion of the polynomial f (z) in powers of z i= h yields f (z)= (2i) [I +(li)h3 Setting h= a (1 i), we get f (z)=(2 i) [1 whence I f (z) 5 (11 4a3I+ 4a3 jif 1/ 5)<V5 4a3 + 4a3 4 +2i 5 a 4 +2i
5
2 a
)
(2k 1) n 4
p < 1r8 ;
4 +2i +31 5 h4+ 1 5 1751.
for 0 <a <- 2 576. Representing the polynomial in the form f (z)= f (zo) {1 + r (cos + i sin p) z0 )k [1 + (z zo) 4 (z)] } cp set z zo= p (cos ID+ i sin ID), take ID= 2mrc and take p so small that (z zo) (z) I < 1. Then I f (z) I= I f (zo ) 11 1 + rpk + rplc (z z0) 4' (z) I> f (zo). 577. The proof is like that for a polynomial, with use made of Taylor's formula for a fractional rational function (Problem 566) which should be terminated after the first term with nonzero coefficient, not counting f (x0). 578. Denote by M the greatest lower bound of I f (z) I as z varies in the region under consideration. By dividing the region into parts, we prove the existence of a point zo, in any neighbourhood of which the greatest lower bound of I f (z) I is equal to M. If necessary, cancel from the fraction the highest possible power of z cp (z) After the cancellation, let f (z)= 4.1 (z) . Then i(zo) 0, for otherwise, in a sufficiently small neighbourhood of zo, I f (z) I would be arbitrarily great and the lower bound of I f (z) I could not be equal to M in a sufficiently small neighbourhood of zo. Consequently, f (z) is continuous for z= zoalso by virtue of the continuity of I f (zo) I= M, which completes the proof.
225
579. The lemma on the increase of the modulus is lacking. 580. Under the hypothesis, a) = f (a) 0, f' ( =f(k-i) (a)=O, pk) (a) (,)
and by Taylor's formula, pk) (a) (za)k [1 + cp (z)], cp (a)= 0. f (z)=P61) -F - k Set 1 f (k) (a) " r (cos p+ i sin cp), z a = p (cos 0+i sin 0). f (a) kl Take p so small that I p (z)I < 1, rpk < 1. Then (z)1= I f (a)I 1 1 +rpk [cos (cp i-k0)+ i sin (cp + k0)] +rpk AI, where IA <1. For 0= For 0= (2m-1)Tc-9,
2m7cp
m=1, 2, ..., k,I f (z)I <I f (a) I. 2 , k , f (z) > f (a)
m=1,
Thus, as 0 varies from k to k +2n, the function I f (z) I I f(a) I changes sign 2k times. Since I f (z) I I f (a) I, as a function of 0 is continuous, I f (z) I I f (a) I vanishes 2k times, thus completing the proof. 581. As in Problem 580, show that Re (1 (z)) Re (1(a))and Im (1 (z)) Im (1 (a)) for z= p (cos 0+ i sin 0) changes sign 2k times as O varies through 1 , 2n, provided only that p is sufficiently small. Setting IT f (a)= r (cos cp+ + i sin p), we obtain, by Taylor's formula, f (z) f (a)= rpk [cos (9 + k0)+ +i sin (cp + k0)] [1 +cp (z)], cp (a)=0. Choosing p so that I cp (z) I <1, we obtain the following, setting 9 (z)=p1(z)+ic p (z): Ref (z)) Re( f (a))= rpk [cos (cp +k0) (1 +91(z)) sin (9 + k0) cp2 (z)], Im(f (z)) Im (f (a))= rek [sin (9+k0) (1 +cpi (z))+cos (cp + ke) cp2(z)]. Putting cp +k0=mn, m=0, 1, 2, ..., 2k, we get Re (f (z))Re (f (a)) = rpk (-1)171(I +E,,,) where cmis the corresponding value of 91(z), I cmI < 1. Whence it follows that Re (1 (z)) Re (1(a)) changes sign 2m times as z traverses the circle I z a I= p. A similar result is obtained for Im (1 (z)) Im (f(a)) by putting 9 + k0= +m1t, m=0, 1, ..., 2k. 582. (a) (x-1) (x-2) (x-3); (b) (x-1 i) (x-1 +i) (x+ 1 i) (x+ 1 +i);
8. 1215
226
(c) (x+ I
1/ V2+1 2-1 ) ( 1/ V2+1 -1 1/ V x+ I 2 2 2 +/ -1/ -1 2 (x+14" 24.1 +i 1/1/ 2-1 2 x(x+1+1/ 1 2+1 2
I1 )
.
(d) (x-V3--V 2) (x -1/3+1/ 2)(x+1/ 3-V 2)(x+1/ 3+V 2).

583. (a) 2nn
k=1
[1 (x _cos (2k-1) 2n
(0+(2k-1) n
IT
\
J (2k-1)n 4m
(b) 2 n x+
(2k-1) n I sin k=1 2n 584. (a) (x2+ 2x+ 2) (x2- 2x + 2); (b) (x2+3) (x2 + 3x + 3) (xs- 3x + 3);
2n
'
c)
k=
(x tans
-
2+1 ) (c) (x2 +2x+ 1+V2 -2 (x+ 1) -V 1/ 2 x (x2 + 2x + 1+1/2+ 2 (x + 1)

n-1 2n
1/ V 22+
(d)fl (x2-2 V 2xcos

k=0
(8k+ 4n1) n
(e) (x2-x Va+2 + 1) (x2 + x V a+2+ 1);
n-I 1) 27t -I- 1) (f) 1 1 (xa - 2x cos (3k + 3n

k=0
-(34+ 170x2+ (23 + I7i)x-(6 +6i); (b) (x+1)3(x-3) (x-4)= x5-4x4-6x2+ 16x2 + 29x + 12 ; (c) (x-i)2(x+ 1 + i)=x2+(1 - Ox2+ (I -2i)x-1- i.
586. 1
k=i
585. (a) (x - 1)2(x -2)(x - 3)(x - 1 - i)=x5-(8 + Ox4+ (24 +70x2
xk (X).
587. (a) (x - 1)2(x -2)(x - 3)(x2-2x + 2)=x6- 9,0+ 33x4 -65,0 +74x8-46x+ 12; (b) (x2-4x+ 13)2 =x6 - 12x5+ 87x4- 376x8 +1131x2-2028x+ 2197 ; (c) (x2+1)2(x2+2x+2)=x6+2x5+4x4+4x2+5x2+2x+2.
227
588. (a) (x-1)2(x +2); (b) (x + 1)2 (x2 +1); (c) (x-1)8. 589. x-4-1, where d is the greatest common divisor of m and n. , and 7 7 /are odd; 1 if at least one of them is 590. xd + ad if the numbers 1 even; d denotes the greatest common divisor of m and n. 591. (a) (x 1)2 (x + I), (b) (x-1)3(x+ 1), (c) 1 (d is the greatest common divisor of m and n). (x0) 592. Denote A0= u v (x0) and factor f (x) into linear factors : f (x)= =(x A0) (x ...(x Ak_ D. Then Xi Xo for j00. Furthermore, f u v( (x x) \ ) r v (u (x)xo v (x)) (u(x)-4_,v (9).
By virtue of the hypothesis and also of the fact that u (x0) Xi v (x0= v (x0) (X0 Ai) 00, the polynomial u (x) v (x) has xoas root of multiplicity k> 1. Hence, u' (x)A0 v' (x) has x0as a root of multiplicity k 1. Furthermore, f
(x) \
(x)
1
[v (x)ik
(x)
(x))
. . . (u' (x)-xk _,
(x))
All u' (x) v' (x), .100, do not, obviously, vanish for x =x0. Consequently, has xoas a root of multiplicity k 1, which is what we set out to v' (x) J prove. 593. If w is a root of the polynomial x2+ x +1, then 0=1. Hence, warn+ wa =0. +wsn+14.0p+2= 1 +, 594. The root X of the polynomial x8 x+ 1 satisfies the equation A3= 1. Hence, X3mVmfi+X3P+2 =(-1)m(-1)"X+(-1)P A2
= (-1)m ( -1)P A [( -1)P (
u' (x)
This expression can equal zero provided only that ( =(-1)P---( I)", that is, if m, n, p are simultaneously even or simultaneously odd. 1 595. x4+x2 + 1= (x2+x +1) (x2 x + 1). These factors are relatively prime, x2 + x + 1 is always a divisor of xam +x34+ 3+x3P+ a (Problem 593). It remains to find out when divisibility by x2 x+ 1 occurs. Substitution of the root X of this polynomial yields ( on a+( 1)p A2=(-1)m(-1)PX R-1)n+(-1)Pj.
This will yield 0 provided only that (-1)m= (-1)"= ( On, that is, if the numbers m, p and n+ 1 are simultaneously even or odd. 596. If m is not divisible by 3. 597. All roots of the polynomial Xk-1Xk-14- ...-F I are kth roots of 1. ka -1-k-1 _i_kaz+1+ Hence, k =1+ +...+ 0, whence follows the divisibility, since all roots of Xk-1+ ... 1 are prime. 598. Substitution of the root w of the polynomial x2-1-x+ I into f (x)= )02 -1. But 1 + w= wa= X, which is a = (I + xm 1 yields (1 + primitive sixth root of unity. Furthermore, w= V, whence f (w)=Am 1.
8*
228 For m = 6n
f (w)= -1 00 , m=6n+1 f (w)=A-A2-1=0, m=6n+2 f (w)=Aa +A- 1 , m=6n+3 f (w)= -3 00, m=6n+4 f(w)= -A+A3-1 00, m=6n+5 f (w)- -X2+)s - 1 =0. Divisibility of f (x) by x2+x+ 1 occurs when m= 6n + 1 and m = 6n + 5. 600. f (w)= m(1+ w)m - mwm- 1= m[A'n-- X3(ni-1)], m=6n+ 1.
599. For m=6n+2 and m=6n+4.
r(ii)=0
only for
601. For m=6n+4. 602. No, because the first and second derivatives do not vanish at the same time. 603. For x= k,
f (k)=1 k 1 +
k (k -1)
1 2
... +(-1)k
k (k -1) . . . 1 -(1 1)k=0. 1 2 ... k
Consequently, the polynomial is divisible by (x-1) (x -2). . .(x - n). A comparison of the leading coefficients yields
f (x)=
r(_ n1 (x 1) (x -2) ... (x-n). )n
fi (1)=0, f2 (1)=0.
604. For m relatively prime to n. 605. If f (xn) is divisible by x-1, then f (1) =0, and, hence, f (x) is divisible by x-1, whence it follows that f (xn) is divisible by xn- I. 606. If F (x)=f (xn) is divisible by (x- a)k , then F' (x)= f ' (xn)nxn -1is divisible by (x-a)k -1, whence it follows that f ' (xn) is divisible by (x- a)k -1. In the same way, f " (xn) is divisible by (x-a)k ...,f (k -')(xn) is divisible by -1) (an) = 0 x- a. From the foregoing we conclude that f (an)=f' (an)= and, hence, f (x) is divisible by (x-an)k, f (xn) is divisible by (xn - an)k. 607. If F (x)= (x8)+ xf2 (x3) is divisible by x2+x+ I, then F(w)=f1(1)+ + wf2(1) =0 (w is a root of x2+x+ 1) and F (w9= (1)+ w2f2 (1)=0, whence
608. The polynomial f (x) has no real roots of odd multiplicity, for otherwise it would change sign. Hence, f (x)= (x)l 2f2 (x), where f2 (x) is a polynomial without real roots. Separate the complex roots of the polynomial f2 into two groups, putting conjugate roots in different groups. The 'products of the linear factors corresponding to the roots of each group form polynomials with conjugate coefficients (Pi (x)+ 2(x) and qh (x)- 2(x). Hence, fa (x)= qi? (x)+M (x) and f (X) = 609. (a) -x1, - x2, ...,
(b) , . , 1
X1 X2
q)ir + (A 'P2)2. 1 1
xn
(c) xi-a, x2-a, ..., xn-a, (d) bx1, bx2, 8r= 4pq - p2.
bxn.
610. One of the roots must be equal to - 2. The desired [relation is
229
1 1 611. x1= 6 , xa= 2 , x3 =
F612. a3-4ab+8c=0. S 13. The relationship among the roots is preserved for arbitrary a. Taking
a =- 4' we get y4+a'y3+b'ya+c'y+d'=0, a'=0, a's -4a'b'+8c'=0, for the transformed equation, whence, c'= 0. 614. clad= c2. d +a (x+ )+ b=0. 615. Division by x2yields x2+ ax x2
Making the substitution x+=z, we get x2 + ax
-z
=x2+
a2x2
- z2
we get the quadratic equation z2+az+b-2 - =0. a a , whence, for z, z, it is easy to find x (generalised reciprocal equations). Having found 616. (a) x=1 1/3, 1 i (c) x= (b) x= + 2i, -2+ i;
2
n c
-1+1/ 5 -1+0/1i
,
; (d) x=1+ 1/ 3
-3+157
'
617. A= +6. 618. (1) b= c=0, (any a),(2) a=-1, b=-1, c=1. 619. (1) a=b=c=0; (2) a=1, b= -2, c=0; (3) a=1, b= -1, c= -1; 2-X2 c= A , where As - + 2 = O. (4) b=A, a=
620. X= -3.
621. q3 +pq+q=0. 622. a?- 2a2.

623. xi=
a,
2i- n- 1 h, i=1, 2, ..., n where 2

h= -
II
12 (n_ I) 4- 24na2
n2-1
624. If the roots formed an arithmetic progression, then, by the formula of Problem 623, they would be: 1 1 3 (a) - , 2 , ; they indeed satisfy the equation; 5 3 1 1 (b) - 2 - , - , , ; they do not satisfy the equation. 625. Let y=Ax+B be the equation of the desired straight line. Then the roots of the equation x 4+ax3+bx2+cx+d=Ax+B form an arithmetic progression. We find them in accordance with Problem 623:
xi= where
h=
a
4
2i-5 , 3 4 h 1=1 2 2 , ' ,
1 12/9a 24b 1 1 3a2 -8b 15 5 2 V
230 Whence
A c=xix4(x2+x3)+ x2x3(xi+ x4)

= h2) _6a 2
h2) a
2
a' 4ab
8 '
dB=x1 x2 x3x4= 16100(36b I l ag) (4b + a2).

Consequently (23 4ab +8c
A=
,B=d
1600
(36b I la2) (4b+a2)
The intersection points will be real and noncoincident if 3a2-8b>0, that is, if the second derivative of 2(6x2 +3ax+ b) changes sign as x varies along the real axis. el+ 1 626. x4 ax2+1 = 0 where a= cc2 627. (x2 x+ 1)2 a (x2 x)2=0, a= ((0,2 _ ,02
0,11 ,+ 1)3
xi+ ... (xi -x); f (xi)= 2 [(xi 628. f' (xi)= (xixi) ... x2) ... (xi xi_ 2) (xi xi + i) ... (xi + (xi x2) ... (xi 2) (xi xi+2) ... (xi x,,)+ + (xi x2) ... (xi (xi x, +2) ... (xi
n
xn_ 4)]=2f' (x4)

(soi)
/-1
1 (if xs xi). xi
x,
629. It follows directly from Problem 628. 630. Let xi=x1+ (i Oh. Then
f' (xi)=( 1)n (i I)! (n hn 1. 631. (a) x +1, (b) x2 +1, (c) x3+1, (d) xa 2x+ 2, (e) x3 x+ 1, (f) x+3,
(g) x2+x+1, (h) x2-2x1/ 2-1, (i) x+2, (j) 1, (k) 2x2 +x 1, (1) x2+x+1. 632. (a) (x 1) fi (x)+(x+ 2) fa (x)= x2 2, (x)+ (x+ 1)12 (x)=x8+ 1, (b) (c) (3x) fi(x)+(x24x + 4) fa (x)= x2+ 5, (d) (1x2) /I (x)+ (x3+ 2x2 x 1) /2 (x)=x3+ 2, (e) (x2+x+ 1) f1 (x)+ (x3+ 2x2 5x 4) .f2 (x)=3x+2, 2x2 2x x 1 (x)=x 1. (f ) 3 (x) + 3 633. (a) M2 (X)= X, (X)= 3X2 X+ 1 (b) M2 (x)= x 1, Ml(x)= xs + x2 3x 2; x4 2x2 2 (c) M, (x) x2 2+ 3 , (x) 2 2x2+ 3x 2x3 +5x2 6 (d) M2 (x) = 6 (x)= 6 '
231
(e) M2 (x)=3x2 +x- 1, MI. (x)= - 3x3 + 2x2 + x- 2; (f) M2 (X) = X33X24X 2, M1 (x)=x4+6x2 +14x2 +15x+7. 16x3- 53x2 - 37x-23 -16x2 + 37x+ 26 , (x)= 634. (a) Ma (X) = 3 3 (b) M2 (x) = 4 -3x, MI(x)= 1 + 2x + 3x2; (c) M2 (x)=35- 84x + 70x2- 2Ox2, M1 (x)=1 +4x+10x2 +20x3. 635. (a) MI (x)=9x2-26x - 21, M2 (X) = 9X3 44X2 39x-7; (b) (x)=3x2 + 3,0- 7 x + 2, M2 (x) = 3X36X2 X + 2. 636. (a) 4x4- 27x8 + 66x2-65x + 24 ; (b) - 5x' +13x6+27x2 -130x4 +75x3+ 266x2- 440x + 197. 637. N (x)= 1 + T
n x+ n(n +1) x2
(n+m - 2) n(n+1) x'n-1; +...+ 1 2 .,. (m-1) m ( 1m +I ) (1 x)2 M (x) = 1 + (1 - x) + m 2 1 (m + n -2) m(m+1) + + (1 x)" 1 2 ... (n-1) (m+1) (m+ 2) ... (m+n-1) m (m+ 2) ... (m+n-1) x (n - 1)! (n - 2)1 1 m(m+1) (m+3) (m+n-1) x2 (n - 3)! 12 m(m+1) (tn+n -2) xn-1 . (n - 1)1
638. 1. 639. (a) (x + 1)2 (x-2)2, (b) (x + 1)4 (x-4), (c) (x-1)3(x +3)2 (x -3), (d) (x - 2) (x2- 2x + 2)2, (e) (x3- x2- x - 2)2, (f) (x2 +1)2 (x-1)3, (g) (x4 +x2 +2x2 +x+1)2.
640. (a) f (x)= x + I + x (x-1) (x-2) (x -3);
1 24 (b) f (x)= - x 4+ 4x2- x2-7x+ 5 ; (c) f (x)= 1 +

(x-1)-
(x - 1) (4x- 9)
(x- 1) (4x - 9) (x-4),
232 f(2)=1
389 =1. 4116 ... (1/=1. 4142 ...); 945 (d) f (x) = x8 9 x2 +21x 8. 1 641. (a) y= (x-2) (x 3) (x-4)
-
1 + (x-1) (x-3) (x-4)-2 (x 1) (x 2) (x-4) 2 4 65 1 + (x 1) (x-2) (x 3)= sx 3+ 10x2 x+15; 3 2 (b) y= 1 [5(1-0xx2 (1+i)X3].
n-1
642. f (x)=
n +1 1 icot krc 2 2 E (1 n
k=1 n-1
Xk
Solution. f (x)=
s=0
(s+i)(xn_i)
(
x_e3)ne! I (s+i)(1_xl 1
n-1 xk n-1 n-1
=1
n-1
s=0
n-1
L (s+ 0 xk .--ks 1 1xe-1 s n L-I s=0 k =0

1 (s+_ks=_ 0 1 n
n-1
1 =_
n
1
E
s=0
(s + 1)
n-1
k=0 n-1
s=0 n-1
Xk
-F
E
k =1
1 n E ( s o ck s= 2
s=0 n-1
Xk
v
k=1
Xk
1 ew1
k
n +1 = 2
n
1 E (1_icotr /11 )
k =1
643. f(x)=
k =1
Yk (Xn 1) = 1
E n
Yk ( 1 x in)
1 xci
k) "nk n =1
f (0)=n
k=1
Yk
644. Set cp (x)=(xx1) -x0 ..(x - x). Let f (x) be an arbitrary polynomial of degree not higher than n-1, let Yh Ya, ynbe its values for x=x1, x2, ..., xn. Then f (x0=
Yi+ Ya+ +Yn
V
k=1
(xi) (X0 d
Yky(xo)
233
Since yl, y2,
y,, are arbitrary,

(xo)
(xk) (xo -xk)
We consider the polynomial

F (x)= n [cp (x0) -p (x)l -(xo- x) cp'_(x).
Its degree is less than n and it vanishes for x=x1, x2, ..., x,,. Hence, F (x)=0. Expand p(x) in powers of (x- 4):
cp (x)=
E
k=0
Ck (X X0)k
We have
E
k=1
(n-
ck (x xak=O. Consequently, ci=- c2 =

1 co. cP (X) =(X Xo)n +Co, Xi= X0+1
A comparison of the coefficients of xn-1 645. x s = 1 ,4 (x - xi) cp' (xi) i=i yields
n
(x)
646. xn-l=
Xn-1yields
1=1 x7-1 (x)

(x
E
n
p' (xi) -0.
9' (xi)
A comparison of the coefficients of
xr1
i=i
n
cr (xi) 1
- 1.
x1 ... x7 -1
X2
. . .
647. ai= E
k=1
Yk ski
where A=
1 1
Xr
... Xn n-1
Oki is the cofactor of the element of the kth row and (1+1)th column of the determinant A.
n-1 n n-1
f
=
ai xi =.
yic
4ki xi=
yk k
k=1
i=A
k=1
where Akis the determinant obtained from A by substituting 1, x, ...,xn-1
for the elements of the kth row,
234
Evaluating the determinants A k and 0 as Vandermonde determinants yields Ak (x xi) . . (x Xk 1) (-X 'c +1) (X Xn) A (xk x1) . . . (xk Xki.)(Xk Xk+i) (xi( cp (x) = (x xk) cp' xk ) where p(x)=(xx1) (xx2)...(xx").
(
(xo, which is what we set out to prove. xk p)(4(:) Whence f (x) E (x _Yk
1 x + x(x-1)+ ...+ x(x-1) (xn+1) 1! 2! n! + (a 1) x + (a 1)2x (x 1) 649. f (x)= 1 1 2 1 (a 1)4 x (x 1) . . . (x n+ 1) . . . n! 2x 2x (2x-2) 2x (2x 2) . . . (2x 4n 4- 2) 650. f (x)= 1 + +. . + 1 1 2 (2n)! x1 (x 1) (x 2) 651. f (x)= 1 21 3! (x 1)(x-2) (x n + 1) _ +(_
648. f(x)=
n1
nl (1 x) (2x) ... (n x) nl x 652. f (x)= cp (a) w (x) p (a) (x a) where cp (x)=(x x1) ( )C x m
1
C x2) (-) 12).
653. We seek f (x) in the form f (x)= Ao +A1 + A2 (x m) (x m 1) 12 (x m) (x m 1) ... (x m n + 1) +... + n

nl
where m, m + 1, . . m+ n are integral values of x for which, by hypothesis, f (x) assumes integral values. Successively setting x= m, m + 1, . m+ n, we get equations for determining Ao, A1, A:
Ao= f (m),
Ak=f (m + k) A0
1
A1
k (k 1)
A2 -kAk- 1.
k = 1, 2, ..., n from which it follows that all coefficients Ak are integral. For integral values of x, all terms of f (x) become binomial coefficients with integral factors Ak and for this reason are integers. Hence, f (x) assumes integral values for integral values of x; this completes the proof.
235
654. The polynomial F (x)= -f (x2) of degree 2n takes on integral values for 2n+ 1 values of x= n, (n-1), ..., 1, 0, 1, ..., n and, by virtue of the preceding problem, assumes integral values for all integral values of x.
3 (x+ 2) + 3)' 1 1 1 1 (b) + + ' 6 (x-4) ' 6 (x-1) 2 (x 2) 2 (x 3) 2+1 2i 2 (c) . + (xi) + 2 (x+i) ' 2 x-1 i 1 1 i (d) + 4 (x-1) 4 (x+ 1) 4 (xi) 4 (x + i) '
E2 I 1( 1 c i V + + ) e = 2 +2 ' 3 x-1 xe x e" 1i 1 f 1+i 1i + 1+i + + (0 16 x-1i x-1+1 x+1i x+1+i)'
655.
(a) 12 (x-1)
(e)
n-1
(g)
ek
k=0
x ek
ilk
ek= COS
2kn
+i sin
2krc
, (h) n Z.-1 x--k

k=1 ck
on- k
n i V
iik =cos
(2k 1) 7T (2k 1) it +i sin n n

on- k Cr4:k
(i)E
k=0
n x k
;
k= n sin 2k-1 it
xk
ok _
(k)
2n 2k-1 X COS n k=1 2n x +2 1 656. (a) 3 (x2+x+ 1) ' 3 (x-1) 1 1 1 (b) 8 (x+ 2) 2 (x2 +4) 8 (x-2) x-2 1 x+2 1 (c) 8 x2 +2x+2 8 x2 -2x+2 ' 2 \ 1 1 (d) 18 ( x2 +3x+3 x2 -3x+3 x2 +3 J n x cos 2k (m+ 1) rc cos 2kmTC 2n +1 2n+ 1 1 [ 1 +2 (e) 2krr 2n +1 x-1 x2 -2x cos 1 k=1 2n+1 +
236
((f) 2n + 1
l x + 1 +2 E
k=1
x COS
2k (m + 1) TC 2kmrc + cos 2n+1 2n + 1 2krc x2+ 2x cos +1 2n+ 1 kit -, x cos n x2- 2x cos +1 x +1
n-1
1 (g) 2n
1 - 1 +2 x +1 x-1 (2k-1) mrc n

n
k=1 X cos
() h 1 -.
E
k=1
cos
(i)
1 +2 (n1)2 x
k=i
(2k - 1) (2m+ 1) Tr 2n (2k 1) rc x2 2x cos +1 2n (_ ok x
(n + k) I (n- k) 1 (x2+ k2) 1 4 (x+ 1)2 ;
657. (a) (b) (c) -
1 4 (x-1)2
1 4 (x + 1) 3 (x -1)8
[n-1
1 1 1 + + 4 (x - 1)2 4 (x - 1) 4 (x+ 1)2; 1 4 (x -1)' + x -1 (x:ick)2 (n 1) 11

k=0
1 (x + 1)2
1 2 x + 1 + x -2 ;
1 (d) n2
k=0
xckEk
217c 2lar ek = cos , +i sin -n n
n
1
n (n + 1) xm- 2
n(n+1) ... (n+m -2)

1 2 . .. (m-1)
X
1.2
m m (m+ 1) 1 1 .2 1 + (1 -x)n + (1 -x)n-1+ (1-x)n-2
m (m+ 1) . . . (m +n -2)
n1 2 ... (n-1) 1 -x
(2 a)n k
(f)
1 (-4a2)"
n(n+1)
kI
(n+k-1)
k=0
I (a
1
X)n k +
1
(a+ X)n
k
CH. 5. POLYNOMIALS AND FUNCTIONS OF ONE VARIABLE n-1
237
(g)(4a2)"
n
E
k=0
(2ar -k
n (n+ 1) . . .(n+ k 1)
k!
x
1
X)"
[(a ix)n- k +( a +
(h)
k=1
g (X k)
+7
k=1
g' (Xic) f ' (xk)g (xk)f" (xk)
[f' (xk)12(xxk)2 L1
[f' (xk)P (xxk)
658. (a) (b) (c) 1 +
1 x-1 x+ 1 + 4 (x+1) 4(x8+1) + 2(x2+1). ,

7 3 3 6x+2 1 3x+2 3 1 1
x x+ x+ 1
1
(x+ 1)2 x 2 +x+ 1 (x2 +x+ 1)2 ;
16(x-1)2 -16 (x-1) + 16 (x+ 1)2 + 16(x+ 1) 4(x8+1) 2n 1 2n-1 1 x-1 + x+ 1 j 4 (x2-1-1)11
(d)
1
One
1
ni
(x-1)2 + (x + 1)2
sin2 1 " c(1 2x cos --L 1n n+ 1)2 (x2 2x cos k

n
1
"-1
+ -
n2
k=--1
n sina
+
659. (a)
cp' (x)
n2 n9
k=1
krc cos (n 1 x 2 n for X 2 2x cos n +,

kn
(x)
(b) xcp' (x)ncp (x) , (c) p (x) ci2 (2) + (1? (1 = (I) 1 17 5
(X)]2p (x) p" (x)
[p (x)]2
660. (a) 9, (b)

661. 0.51x+2.04.
(c) 17.
662. y= 77- [0 .55x2+2.35x + 6.98].
1 into f (x), we obtain, after multiplication by q", 663. Substituting .1 nopn+non-r q+ whence
ao Pn = (aipn-2+
an_l pqn-1 +anqn =
an_ Ipqn -2 + an qn-1),

pri-2 q+
an qn _ (a pn- +
238
The right sides of these equations contain integers. The numbers p and q are relatively prime. Hence, aois divisible by q, and anis divisible by p. Now arrange f(x) in powers of xm: f(x)=a0(x m)n+ci (xm)n-i+ . + cn_1(x m)+ cn. The coefficients c1, c2, cnare integers, since m is an integer, cn=f(m). Substituting x= -11, we get
ao(p mq)n+cL (pmq)n-lq+ .

n qn
qn + en qn = 0
whence we conclude that pmqis an integer. P pmq p m is in lowest terms, the numbers Since the fraction
p mq and q are relatively prime. Hence, cn = f(m) is divisible by p mq,

which is what we set out to prove. 664. We give a detailed solution of (a). Possible values for p: 1, 1, 2, 2, 7, 7, 14, 14. Only 1 for q (we take the sign to be attached to the numerator). f (1)= 4. Hence, p 1 must be a divisor of 4. We reject the possibilities p=1, 2, 7, 7, 14, 14. It remains to test 1 and 2 f (-1)00, f (2)=0. The only rational root is x1=2. 1 (b) xi= 3; (c) xi = 2, x2=3; (d) x1= 3, x2=2 ' 2 3 5 3 1 ; , , , -71 ; (f) 1, 2, 3; (g) (e) (h) no rational roots; (i) 1, 2, 3, +4; (j) 1 ; (k) xi= x2= 1 (1) x1= x2= 1 , x3=x4=-3; (m) x1=3, x2=x3=x4=x5= 1; (n) x1 =x2 =x2 =2.
665. According to Problem 663, p and pq are odd at the same time. Hence, q is even and cannot equal unity. 666. By Problem 663, px1q= +1,px2q= +1, whence (x2 x1)q= 2 or 0. The value 0 is dropped because q> 0, x2 Ox1. Putting x2 >x1for definiteness, we get (x2 x1)q=2. This equation is impossible for x2 x1> 2. Now put x2 x1=1 or 2. The only possible values for p and q, for which equation 2 (x2 x1)q= 2 is possible, are p=x1q+1, q= , whence the sole possi./C2 .X1 1 xi-Ex2 . The proof is complete. bility for a rational root =xi+ q 2 667. The Eisenstein criterion holds: (a) for p= 2, (b) for p= 3, (c) for p=3 after expanding the polynomial in powers of x-1.
668. Xp(x)= (x 1)P -i + 4 / - (x-1)P-2 + P (iP .21)(x 1)P- +

All coefficients C k --
+p.
, k,-..p 1, are divisible by 2 ... k p because k! Ck=p (p-1) ... (p k+1) is divisible by p, and k! is relatively prime to p. Thus, after the expansion of X, (x) in powers of x-1, the Eisenstein criterion holds for Xp(x) for p prime.
1
p (p 1) ... (p Ic+1)
239
669. Apply the Eisenstein criterion for the number p, setting x= y+1 :
X k (X) = =
(y+ o pk-i_l
(y + 1)Pk -1
The leading coefficient of the polynomial cp is equal to 1. The constant term of cp (y), equal to cp (0)=. X pk (1)=p, is divisible by p and is not divisible by p2. It remains to prove that the remaining coefficients are divisible by p. To do this, we prove by induction that all the coefficients of the polynomial (y+ OP" -1, except the leading coefficient, are divisible by p. This is true for n= 1. Suppose it is true for the exponent pn -1, that is, (y+ 1)Pn 1 = y Pn + 1+ +pwn_ i(y) where w,1(y) is a polynomial with integral coefficients. Then (Y+1)Pn = Y ( Pn-l + +Pwn-1.(Y))P=U Pn-1 + +P4) (Y)=Y Pn 1 Pwn(Y); 4)(y) and w. (y) are polynomials with integral coefficients. Thus,
k Y P +PWk (y) YP
k-1 -Y
k 1 Pk P
+p
Wk
+PWk -. 1(Y)
kP k YP k (y) k YP +PW k 1 (y)
-FPX(Y). The coefficients of the polynomial x (y) are integral, since x (y) is the quotient obtained in the division of polynomials with integral coefficients, and the leading coefficient of the divisor is equal to unity. Hence, all coefficients of the polynomial cp (y), except the leading coefficient, are divisible by p. The conditions of the Eisenstein theorem are fulfilled. 670. Suppose the polynomial is reducible: f (x)=cp (x) (x). Then both factors have integral coefficients and their degrees are greater than 1, since f (x) does not have, by hypothesis, rational roots. Let
=Y
P k P k 1
i xk 1+ . + (x)= bo Xk b -co xm+cixm-1+ ... +cm (I) (k)=k m k+ m = n. Since bkc,,, = anis divisible by p and is not divisible by p1, we can take it that bk is divisible by p, cmis not divisible by p. Let bibe the first coefficient of cp (x) from the end that is not divisible by p, i>0. Such exists since ao=boco is not divisible by p. Then ami_ i= bicm +1 ,1+ m_ 1+ ... is not divisible by p, since bicmis not divisible by p, and bi bi+2, ... are divisible by p. This contradicts the hypothesis because m+i>2. 671. Factoring f (x) into irreducible factors with integral coefficients we consider the irreducible factor 9 (x), the constant term of which is divisible by p. Such exists since an is divisible by p. We denote the quotient after the division of f(x) by p(x), by 4(x). Let
(x) = b 0 ez + bi x"1-'+
+bm,
(1) (x)=coxh+
+ . . . + ch
and bibe the first (from the end) coefficient of 9 (x) not divisible by p; ch is not divisible by p since a.= bmch is not divisible by p2. For this reason ah+t =bich+bi}1ch_ 1+... is not divisible by p, whence follows h+i.k. Consequently, m>m+h+i-k=n+i-k>n-k. f (-1)= -1. 672. (a) f(0)=1, f (1)= If f (x)=cp (x) 4i(x) and the degree of cp (x):.c. 2, then cp (0) = 1, 9 (1)= 1, cp (- 1)= +1, that is, p (x) is represented by one of the tables:
240
(x)
-1
0
1 1 1
1 1 I
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
The last 5 tables may be omitted since the last 4 define polynomials that differ only in sign from the polynomials represented by the first four tables, and the fourth defines a polynomial identically equal to unity. The first three yield the following possibilities: p (x)= (x2+ x 1), p(x)=x2 x 1, cp (x)=2x2 I. Tests by means of division yield f (x)= (x3+ x 1) (x2 x 1). (b) Irreducible, (c) irreducible, (d) (x2 x I) (x2-2). 673. A reducible polynomial of degree three has a linear factor with rational coefficients and therefore has a rational root. 674. The polynomial x4+ axa + bx2+ cx+d has no rational roots and can be factored (in case of reducibility) only into quadratic factors with integral coefficients: x4+ax3-1-bx2+ex+d=(x2+)x+m) (x3+ [Lx+ n). The number m must obviously be a divisor of d; inn= d. A comparison of the coefficients of x3and x yields
X-1-11= a, nX+ my.= c.

This system is indeterminate only if m=n, c= am, that is, if c2 = a2d (see Problem 614). But if m On, then X =
c am cm am2 and this completes the proof. nm d ma
675. In case of reducibility, it is necessary that
x6-Fax4+bx3+cxs+cbc+e=(x2+Xic+m) (x3+Vx2+X"x+n).
The coefficients of the factors must be integral. A comparison of the coefficients yields mn= e, whence it follows that m is a divisor of e. Furthermore, =a, nX + mX" =d,
m+XX' +X" =b, n+XX"+ mX' =c mX" =d an, X (mX" nX')+ maX' nX" = cm bn and, consequently, (d an) X + m2X' nX" = cm bn. Solving this equation and X+ ?s '= a, nX+mX"=d simultaneously, we get atria cm2 dn+ be X m3 n2+ ae dm
which completes the proof. whence
241
676. (a) (.70 2x + 3) (x2 x 3), (b) irreducible, (c) (x2 x-4) (x2 + 5x + 3), (d) (x2 2x + 2) (x2 + 3x + 3). 677. Without loss of generality, we can seek conditions under which x4+ + px2+ q can be factored into quadratic factors with rational coefficients, because if the polynomial has a rational root x1, then x1will also be a rational root and the linear factors corresponding to it can be combined. Let x4+px2+q= (x2+Xix + Then
(x2+X2x+ 112).
1 =0, 1 Al+ X2 =0, )\11-t2+ X21 [Ai+ Ai X2 + tta =p,
q.
If X1=0, then A2=0 too. In this case, for the existence of rational [L1and z it is necessary and sufficient that the discriminant p2-4q be the square of a rational number. Let x1 00. Then A2 = Ai, 42=iand furthermore q= 2[11 p = Thus, for the reducibility of the polynomial x4+ px2 + q it is necessary and sufficient that one of the following two conditions be fulfilled: (a) p2 -4q is the square of a rational number; 2l p is the square of the (b) q is the square of the rational number rational number Al. 678. If x4+ axa + bx2+ cx+ d= (x2+ p + q1) (x2+ P2x + 90, then, since Pi+ P2= a, we can write
1 x4+ ax3+ bx2+ cx+ d = (x2+ ax+

2
)2 ( PlP2 x+ q12412 ) 2
2
where X= qi+ q2. Whence it follows that the auxiliary cubic equation has the rational root X= qi+ qz. 679. Let f (x)=cp (x) (1) (x) and cp (x), 4,(x) have integral coefficients. Since f (ai)= 1, it must be true that cp (ai)= I, (1) (a1)= 1 or cp (a)= 1, 4) (ad =1 and, hence, 9 (ad + 4)(ai) = 0 i =1, 2, ..., n. If cp (x) and 4, (x) are both nonconstants, then the degree of p(x) + 4, (x) is less than n, whence it follows that p(x) + 4, (x) is identically zero. Thus, we must have f (x)= [cp (x)]z. This is impossible since the leading coefficient of f (x) is positive. 680. If f (x)=cp (x) 4, (x), then cp (a)= 4,(ai)= 1 since f (a;) = 1 . Hence, if pand 4) are nonconstants, cp (x) is identically equal to 4,(x) and f(x)=[p (x)12. This is only possible for even n. Thus, the only possible factorization is
(x a2) (x a2). . . (x ad+ 1= [cP(x)]a. From this we conclude (considering the leading coefficient of p(x) positive) that
0, (x)+ 1 =(x al) (x (x) 1 =(xaz) (x az)...(x ad.

(In order to have the permission to write these equations, we must change the ordering of the numbers al, az, ..., an.) And, finally, (x a1) (x a_1)(x az) (x az)...(x ad =2.
242
Put al> a3 > > an_ 1. Substituting x=a2k, k =1, 2, ..., 2 - in the latter equation, we get (a3k al) (a2k a3) (a2k a_i) = 2 which is to say that the number 2 must be factorable in inways into 2 =2,
n .
tegral factors arranged in increasing order. This is only possible when
2 = 2 (-1)=1 2, and when =1 These two possibilities lead to the 2 * two cases of reducibility of the polynomial f (x) that are mentioned in the hypothesis of the problem. 681. If an nth-degree polynomial f (x) is reducible for n = 2m or n = 2m + I, then the degree of one of its factors cp (x) does not exceed m. If f (x) assumes the values +1 for more than 2m integral values of the variable, then p (x) also assumes the values I for the same values of the variable. Among these values, there will exist for cp (x) more than m values equal to +1 or 1. But then p(x)= +1 or I identically. 682. The polynomial f (x) has no real roots. Hence, if it is reducible, its factors cp (x) and 4) (x) do not have real roots and therefore do not change sign for real values of x. It may be taken that cp (x)> 0, 4) (x)>0 for all real values of x. Since f (ak)= 1 , it follows that cp (ak)= (1) (ak)= 1, k =1, 2, ..., n. If the degree of cp (x) [or 4)(x)] is less than n, then cp (x) = 1 [or 4)(x) = 1] identically. Hence, the degrees of p (x) and 4)(x) are equal to n. Then cp (x)= 1 + + cc(x (ID . . . (x an), 4)(x)= 1 + (3 (x al) (x a,,), where a and p are some integers. But then f (x)= (x ai)2 (x an)2 + 1 =1 + (a + (3) (x a]) ... (xan)+4 (x- air ... (x an)2. A comparison of the coefficients of x" and xn yields a system of equations, 4=1, + (3=0, that has no integral solutions. Consequently, f (x) is irreducible. 683. Let f (x) assume the value 1 more than three times. Then f (x) 1 has at least four integral roots, i. e.,
f (x)- 1 =(x- al) (x aa) (x a0 (x an) h (x) where al, a2, a3, a4and the coefficients of the polynomial h (x) are integers. For integral values of x, the expression (x a1) (x a3) (x a3) (x a4) is a product of distinct integers. Two of them can be equal to +1 and 1, the remaining two differ from +1. Hence, their product cannot be equal to a prime number, in particular, 2. Thus, f (x) 1 0 2 for integral values of x and, hence, f (x)0 1.
684. Let f (x)=p (x) (1) (x). One of the factors, cp (x), is of degree
and
assumes the values 1 for more than integral values of x. Since 6, 2 2 it follows that +cp (x) or cp (x) takes on the value 1 more than three times and, by virtue of the result of Problem 683, it cannot take on the value 1. Thus, p (x) or cp (x) assumes the value +1 more than
times, and hence,
ep (x) or cp (x) is identically unity. Consequently, f (x) is irreducible.
243
Sharpening the reasoning, we can prove the validity of the result for n 8. 685. Let a [cp (x)]2+ bco (x)+ 1= 4' (x) w (x). One of the factors is of degree n; (1)(x) takes on the values 1 for x= a1, ..., anand since n>7, all these values of 4 (x) must be of like sign. Hence (x- al) a2) (x- ad= 1 +cap (x). (x)= 1 (x+ If a 0, then w (x) also has degree n and w (x)= 1 + Scp (x). But the equation a [cP (x)]2 +bcp (x)+1= 1 + cap (x)] E 1 + [icp (x)] is impossible since the polynomial ax2+ bx+1 is irreducible by hypothesis. 686. (a) f (x)= a, xn (1+ Let max
ak
ao
+...+ " ao x ao x =A. Then for I xl >1
A 11 A I ->0 1a, xn 1 I x .-1 a, xn 1[1 I f (x) I ?. I x 1 11Ix 1 for I x 1>l+A. + V ai r-1 n a, (x ) -2 an 1 -+ -= (b) f (x)= a, (x r pn P P P2 P V' P By virtue of (a), for all roots ak I x1 ak , whence 1x1-<...p+ max 1 +max aopk -1 ao Pk
k
(c) Put p =max
ak . Then a,
ak ak ak max a0 p k -1 a, a Consequently, the moduli of all roots do not exceed p+p= 2p=2 max
k 1
Vl ak
k
T I,
ak
a1
(d) Put p =max I/
ak
a1
. Then 1 a k I
k-1
I al Ipk
a, p k
ao
Hence, the moduli of the roots do not exceed ak +max 1/ ao ao 687 Let f (x)= a, xn+aixh-l+ ...+a,,, (x)= bo xn- xn -1- ...- bn, 0< I ao I, b1? -. I at I, ..., b I an :. Obviously, 11(x) I cp (I x D. b - 2 ... b b bn Furthermore, cp (x) = 60 xn (1 bb x G o x2 x n
p+
a1
244
The expression in the brackets increases from- m to 1 for x varying from

0 to +
Hence, cp (x) has a unique positive root and 9 (x)>0 for Because of this, for I x I > we have I f (x) ( I x I) > 0, whence it follows that the moduli of all roots of f (x) do not exceed 688. (a) Let A =max
ak
. It is obvious that A lx1"+1
A - lao x.i (i xr If (x)1?whence for I x I >1, f (x)1>lao zni (1

A 1x
Ix illn )
xl -0)
I aoxn-r+i. I ( ao x n-r+1 = [I X1r-1(41-1) Al> [(I x I-1)r - A]. Ix 1-1 lxI-1

r
For I x I >1+ 1//1 we have I f(x) I >O. (b) pn f (x) = a 0 ( .) n + ..- ( .)n-r + ... + A pn \ P/ Pr \ P / By virtue of (a), for all roots of f (x) we have
r r
<1 + Vmax k-r (c) Set p = maxi/
ak
ao P k
ak
ar
whence I x 1<p+ Vmax
ak ao pk-r
. Then lak -.1 (41 Pk r and the moduli of all
roots of the polynomial do not exceed

r ar r
7 10
+p=
ar
ao
k-r + max 1/I ak

1
ar
-
689. For negative roots of the polynomial the assertion is obvious. For the sake of definiteness, set ao> 0 and denote cp (x)= aoxn - bixn -1_ 62x"-2 -b", where bk =0 for ak >0, bk = - ak for ak<O. Then, for positive x, it is obvious that f (x) cp (x). Furthermore, 9 (x) has a unique nonnegative root (see Problem 687) and cp (x)>0 for x> Hence, for x> f (x) cp (x)> 0. 690. This follows directly from 688, 689, 686 (c). 692. Expanding f (x) in powers of x - a, we get, for x a,
.
f (x) = f (a)+
f (a) f ' (a) (x - a) + (x-a)2 + 1 2 1
( n)
(a) (X n!
> 0.
693. We obtain the upper bound of the roots by using the results of Problems 690, 692. To determine the lower bound, substitute -x for x:
(a)
0<x;<3, (b) 0 <xi < 1, (c) -11 <xj <11, (d) -6<xj<2.
245
694. (a) f=x3-3x - 1, f1=x2-1, f,=2x+1, f= +1. Three real roots in the intervals (-2, -1), (-1, 0), (I, 2). 2x + 1, f a= +1. Three real (b) f=x8+x2-2x- I, f1=3.x2 +2x- 2, roots in the intervals (-2, -1), (-1, 0), (1, 2). (c) f=x2-7x+7, fi=3.x2-7, f3=2x-3, f 3= +1. Three real roots in the intervals (-4, -3), (1, 3 ) ' 3 2) 2(d) f=x3-x+5, A=3x2 - 1, f2=2x - 15, f a= -1. One real root in the interval (-2, -1). (e) f=x3 +3x -5, fi =x2 +1. One real root in the interval (I, 2). 695. (a) f=x 4-12x2-16x -4, f1=x2- 6x -4, fa =3x2 +6x +2, f3= =x+1, f4=1. Four real roots in the intervals (-3, -2), (-2, -1), (-1, 0), (4, 5). (b) f=x4-x- 1, f1=4x8-1, f2=3x +4, f a= + 1 . Two real roots in the intervals (-1, 0) and (1, 2). (c) f=2x4-8.x8+8.x2-1, f 1=x3-3x2+2x, h=2x2- 4x + 1,1 f a=x- I, 14=1. Four real roots in the intervals (-1, 0), (0, 1), (1, 2), (2, 3). (d)f=x4+x2-I, f1=2x2 +x, 12= -x2 +2, .19 = --Xs _A = -1. Two real roots in the intervals (-1, 0) and (0, 1). (e) f=x4+4x3-12x+9, f1=x2 +3x2 -3, f2 =x2+3x -4, f a= -4x+3, L= 1 . There are no real roots. 696. (a) f=x 4-2x3-4x2 +5x + 5. f1=4.x2- 6x2 - 8x + 5, f2 = 22x2- 22x-45, f8=2x- 1, /4=1. Four real roots in the intervals (I, 2), (2, 3), (-1, 0), (-2, -1). (b) f=x4-2x4+x2-2x +1, A= 2x2-3x2+x-1, fa=x2 +5x-3, fa= = -9x+5, f4= -1. Two real roots in the intervals (0, 1), (1, 2). (c)f=x4-2x2 -3x2 + 2x +1, fi=2x8- 3x2 - 3x + 1,f2= 9x2 - 3x - 5, f3=9x+ +1,1 4= +1. Four real roots in the intervals (-2, -1), (-1, 0), (0, I), (2, 3). (d) f-=x 4-X3 +X2-Xf1=4x3-3x2 +2x -1, fa = - 5x2+ 10x + 17, f 3= -8x-5, f4= - I. Two real roots in the intervals (1, 2), (-1, 0). (e) f=x 4-4x3- 4x2+ 4x + 1, fi=xs - 3x2- 2x + 1, = 5x2- x - 2, f 3=18x + +1,1 4= +1. Four real roots in the intervals (-2, -1), (-1, 0), (0, 1), (4, 5). 697.(a) f=x4- 2x3- 7x2+ 8x + 1, A= 2x2- 3x2- 7x +4, f, = I 7x2- 17x-8, f9=2x-1,1 4=1. Four real roots in the intervals (-3, -2), (-1, 0), (1, 2), (3, 4). (b) f=x 4-4x2+x+1, fi=4x4-8x +1, = 8x2- 3x-4, h=87x-28, .f4= + 1. Four real roots in the intervals (-3, -2), (-1, 0), (0, 1),(1, 2), (c) f=x 4-x3-x2-x+1, fi=4x3- 3x2-2x-1, f2 =11x2 +14x -15, fa= = -8x + 7, f4 = -1. Two real roots in the intervals (0, 1) and (I, 2). (d) f=x4 - 4x3+ 8x2- 12x + 8, fi=x8- 3x2 +4x - 3, fa = -x2+ 5x- 5, 4= - 1 . Two real roots x1=2, 1 < x2 < 2. fa= -9x +13,1 (e) f=x 4-x3-2x + 1, fi=4x3-3x2 - 2, f2 =3x2 + 24x- 14, f3= - 56x + 31. f -1. Two real roots in the intervals (0, 1) and (1, 2). 698. (a) f=x 4-6X2-4x + 2, f 1=x2 - 3x - 1, f2 = 3x2+3x -2, f 3=4x+5, 1 4=1. Four real roots in the intervals ( -2, -2 ' - 3 -1), (0, 1), (2, 3). (b) f=4x4-12x2+8x-1, fi= 2;0- 3x +1, + 1, fa = 2x-1,
f4= 1. Four real roots in the intervals (-3, -2), (0, and (1, 2).
, 1)
246
(c) f=3x 4+12x3+ 9x' -1, f1=2x3+ 6x2+ 3x , f2=9x2+9x + 2, f 3=13x +8, , , , f4= 1. Four real roots in the intervals (-4, -3), ( -1, (0, 1). (d) f= x 4 -.7C3-4X2 +4x + 1, f2 =4.x8 - 3x2- 8x + 4, f2 = 7x2 - 8x -4, f8= , 2) , (-2, =4x-5, f4= I. Four real roots in the intervals (1, - I), (- 1, 0). (e) f=9x 4-126x2-252x-140, fi=x3- 7x-7, f2= 9x2 +27x +20, fa= =2x +3, /4 =1. Four real roots in the intervals (4, 5), ( - , -1) , , -s ), (-2, 5 fi =x4-3x2+1, f2=4x8-8x+3, 13= 3 ,
4), (
699. (a) f 2x 5- 10x3+ 10x - 3,
=4.x2 + 3x -4, f4=x, f6=1.Five real roots in the intervals ( -2, (- 3

'
(0, 1 ) (I 1L (1, 2) \ /'\2' / (b) f= x6- 3x5- 3x4+ 11 - 3x2- 3x +1, f1=2x5-5x4-4x3+11x2-2x-1, 26x+ 5, f2 =2x- 1, C3 -X2 +4x - 1, f = 420- 6x2 + I, f4=26x2 f2=3x 4 - 6, 1 6 =1. Six real roots in the intervals (-2, -1), (- 0), (0, , 1), (1, 2), (2, 3). (c) f= x5+ x4-4x3-3x3+3x +1, f1=5x4+4x3-12x2- 6x + 3, f2=4x3+ +3x2- 6x - 2, f = 3X2 +2x-2, f4=2x +1, f6=1.Five real roots in the inter3-1) vals ( -2, - 2)' ' , (-1, 0), (0, 1), (1, 2). (d) f= x5 5x3- 10x2 + 2, fi= x4-3x2 - 4x, fa= x3+3x2- 1, f = -2X2 + +X+ 1, f4= - 3x- 1, f 5= - I. Three real roots in the intervals (-1, 0), (0, 1), (2, 3). 700. (a) f= x4+ 4x2- 1, x, f2=1. Two real roots in the intervals (-1, 0), (0, 1). (b) f= x4- 2x3+ 3x2- 9x +1, f,.= 2x-3, fa=1. Two real roots in the intervals (0, 1) and (2, 3). (c) f= x4- 2x3+ 2x2- 6x +1, f1=2x - 3, fz= l. Two real roots in the intervals (0, 1) and (2, 3). (d) f= x5+ 5x4+ 10x2- 5x -3, fi= x2 +4x - 1, f2=5x-1,fa =1. Three real roots in the intervals (0, 1), (-1, 0), (-6, -5). 701. The Sturm sequence is formed by the polynomials x3+ px + q,3x2+ p, -2px-3q, -4p3-27g8. If -4p3- 27,72 > 0, then p < O. All leading coefficients of the Sturm polynomials are positive and so all the roots of x3-1-px+q are real. If -4p3- 27q2 <0, then, irrespective of the sign of p, the Sturm sequence has, for - cu, two changes of sign, and for + co, one change of sign. In this case, x3+ px+ q has one real root. 702. The Sturm sequence is formed by the polynomials nq r-1 Xn+pX+11, nxn-'+p, -(n-1) px-nq, -p n \ (n-1) p For odd n, the sign of the last expression coincides with the sign of A= = qn-2. If A >0, then necessarily p <0. In this case, the polynomial has three real roots. If A <0, then, irrespective of the sign of p, the polynomial has one real root.
-
247
For even n, the sign of the last expression in the Sturm sequence coincides with the sign of pA, where A = (n On pn nn q'' The distribution of signs in the Sturm sequence is given in the following table for various combinations of the signs of p and A:
.
fl
fa
I. p> 0,
4>0
oo
+co 2. p<O, A >0 co +co 3. p>0, A<0 co +co 4. p<O, i<0 co + co
+ +
From this table it follows that for 0>0 the polynomial has two real roots, for 0<0 there are no real roots. 703. The Sturm sequence is formed by the polynomials f=x5-5ax3+
+5a5x+2b, fi= x4 3ax2+ a2, f 2=ax3-2a2x b, f.= a(a2 x2 bx a3), f 4= =a(a5 b2)x, f5=1. If A = a5 b2>0, then a > 0, and all leading coefficients of the Sturm polynomials are positive. In this case, all five roots of the polynomial f are real. If A <0, then, depending on the sign of a, the distribution of signs looks like
this:
f a> 0 co
+co
fl
fa
f3
f4 f5
a < 0 oo
+co Consequently, for A <0, the polynomial f has one real root. 704. Let fx and fx+Ibe two consecutive polynomials of a "complete" Sturm sequence. If their leading coefficients have the same signs, then their values, for + co, do not constitute a change of sign, while the values for co yield a change of sign, since the degree of one of the polynomials is even, while the degree of the other is odd. Now if the leading coefficients have opposite signs, then the values of fx and J, lfor + co yield a change of sign, and for co do not. Therefore, denoting by v1and v3the number of variations of sign in the Sturm sequence, for oo and + oo, we have that vi+ v2 = On the other hand, v1 v2 is equal to the number N of real roots of the polynomial. Consequently, va
nN
2, which is what we set out to prove.
705. This is proved like the Sturm theorem, with the sole difference that we have to see that there is an increase (not a decrease) in the number of variations of-sign per unit when passing through a root of the original polynomial.
248
706. The sequence of polynomials thus constructed is a Sturm sequence for the interval xo -.5x< + co and satisfies the conditions of Problem 705 for the interval - co <x5xo. Hence, the number of roots off in the interval (xo, co) is equal to v (x0)- v (+ co), the number of roots of f in the interval (- co, xo) is equal to v (x0)-v (- co), where v is the number of variations of sign of the corresponding values of the polynomials. The total number of real roots is equal to 2v (xo)- v (+ oo)- v (- co). 707. Applying the Euler theorem to x2 2
Pn=(-lr e
dn e
x2
x2
(- e
2)
dxn
=(-1)ne2 x'
dxn-x1
x,
yields
x' Pn=(-1)"--1 eT
X
dn -i e 2 +(n dxn 1
-
dn-2
-
dxn -2
whence
P=xP_ 2-(n-1) Pn_ 2.

On the other hand, differentiating the equation defining P_1, we get xa x2 x2 x2 2 dn ' e T dn +(__ 1)n-1 e
Pn_ 1 =(-1)"-' xe 2
dxn-
dxn
whence
-I= xPn -i -Pn.

Comparing this with the previous formula, we get
' _ 1 = (n -1) Pn _2 and so P' n =n Pn -i Pn
It follows from the derived formulas that the sequence P, Pn _ 2, , n Po=1 is a Sturm sequence for the polynomials P since Pn_1differs from P' in the factor n alone, and PA_ Iis, to within a positive factor, the remainder (taken with sign reversed) after division of PA} , by PA. All the leading coefficients of the polynomials Pnare equal to + 1. Hence, all the roots of P are real. 708. Differentiating the equation defining Pn, we obtain
P,;=(-1)" ex
whence
do (xn e -x)
dx"
- - (-1)" ex
do (nxn-1e-x -xn e -x)
dx" dx"
P,;=(-1)n nex
Furthermore,
d. (xn-1e-x
P,,=(-1)" ex
-
dnn-1e-x)
dxn
do 00-1 e dn-1 (xn -1
-
=( 1)"ex [x
n "_npn_i p
j
dxn
+n
dxn-1
249
whence On the other hand,
xP,;=nPn+ Pn =(- nex

dn -
P n -1.
I) xn- 2 e - x _ xn-i
dxn-
' _ 1 + nPn_ i. Multiplying by x and substituting in place whence Ph = -nP,, of xPn and xPh _ 1their expression in terms of P, Pn_ 1, P,,_ 2, we get Ph = (x- 2n +1) P"_ 1-(n-1)a Pn-2.
From these relations it is seen that consecutive polynomials P,, do not vanish simultaneously, and if Pn_ i =0, then Pn and Pn _ 2 have opposite signs. 1 xPn ' Pn -1changes Furthermore, from Pri-1 n 1.). it follows that Pn Pn sign from minus to plus when going through a positive root of Pn. Thus, the sequence Pn, Pn_1, . P1, Po= 1 is a Sturm sequence for P,, in the interval (0, CO). The leading coefficients of all P,, are equal to unity. P,, (0)= (-1)nnl. Hence, v (0)- v (+ oo)= n, that is P,, has n positive roots. xn 709. En=E,,_ 1. Also, En = En-i- (- ) nl Therefore, the polynomials E,,, En_ 1and - ; x form a Sturm sequence for E on the interval (- oo, - e) for arbitrarily small e. The distribution of signs is given by the following table: _ _ on (_ -e I + +
l)n-1
(- 1)n-1
Hence, for even n, the polynomial Enhas no negative roots, for odd n, the polynomial Enhas one negative root. Furthermore, for x?:-0, the polynomial E,,(x)> O. 710. Use the Euler formula to transform the identity
cln+1 (x2 ex ) dn[(2x-1)e x
]
dxn+2
We get
dx"
I I 1 dn+ 2- e x dn ex do-1 ex x2 +2(n+ 1)x dx" +(n+1) n dxn + 2 dxn- 2

= (2x- 1)
dne x dnle x + 2n dx" dxn-2
whence Pn=(2nx + 1) Pn-l-n (n-1) Pn_ 2 X2. On the other hand, by differentiating the equation defining Pn_1, we get
P,,= (2nx + I) Pn_ 1-x2 Pn _ 1 . Comparing the results, we see that Ph _ i =n (n- 1) P_ 2 and, hence, Pn = (n t 1)nPq _t. By virtue of the established relations, the sequence of polynoT
250
mials P., P_1, Fi,_ 2, ..., Po= 1 forms a Sturm sequence for P.. The leading coefficients of all P. are positive. Consequently, all the roots of P. are real. 711. Computing
do
\ x2+1
do
( 1 dx"
x2+1 I
dx"
by two methods, we obtain P2 -2xP,,,+(x2 +1)P_ 2 =0. Differentiation of the equation defining P,s_1yields P.=2xP_1 P;,_1whence P _ 1 = nP,,_ 2 and, hence, P.' = (n + 1) It follows from the derived relations that P., P,,_ /, ..., P0 =1 form a Sturm sequence for P.. All the leading coefficients of the sequence are positive and so all the roots of P. are real. The solution of this problem is straight forward. Namely,
x2 + 1
1 1 I 1 + 1 x2+1 = 2i k xr x+i
whence we find that
Pn
(x) =
Zl
Rx
+ _ _ +1 krc
k =1, 2, ..., n. It is easy to compute that the roots of P are cot n + 1 , 712. Using the Euler formula to expand the identity
d"
we get
x2 +1 x2+ 1 dxn
do-1
V x2 +
dxn-1
P. (2n 1) xP_ 1+ (n-1)2 (X2 + 1) P._ 2 =0.
Differentiating the equation defining P_ 1, we get Pn (2n 1) xPn_ 1 + (x2 + 1) P _ 1 = 0 whence P.' _ 1= (n 1)2P._ 2 and P;,= n2 P n i. From the relations found, it follows that P, P,, _ 1, ..., Po=1 form a Sturm sequence. Since the leading coefficients are positive, all the roots of P. are real. 713. The functions F (x), F' (x) and [ f' (x))2form a Sturm sequence for F. The leading coefficients of the sequence, 34,124 and 94, are positive. Hence, the number of lost changes of sign when x passes from co to + co is equal to two. If f has a double root, then F has one triple root and one simple root. If j has a triple root, then F has a quadruple root. 714. If some one of the polynomials in the Sturm sequence has a multiple root xoor a complex root a, then this polynomial can be replaced by a polynomial of lower degree by dividing it by the positive quantity (x x0)2or (x a) (xa'). Subsequent polynomials can be replaced by remainders (taken
.
251
with reversed signs) in the Euclidean algorithm for the replaced polynomial and the preceding one. Then the number of variations of sign for x= - oo will be < n -2, where n is the degree of the polynomial. Hence, the number of real roots is surely < n - 2. 715. Let F (x)= (x2- 1)12 . F(x) has -1 and +1 as roots of multiplicity n. F' (x) has -1 and +1 as roots of multiplicity n- 1 and, by Rolle's theorem, one more root in the interval (-1, + 1), F" (x) has -1 and +1 as roots of multiplicity n-2 and two roots in the open interval (-1, +1) and so on. (x)= Pn (x) has n roots in the open interval (- 1, 1). 716. Let xi, xk be distinct roots of f (x) of multiplicity al, a2, cck, f ' (x) < The function cp (x)= < x2 < xk. is continuous in the open x1< x2
f (x)
intervals (- 00, x1), (x1,x2), (xk_ 1, x k) and (xk, + co) and ranges from 0 co, x1), from + co to - co in each of the intervals to - co in the interval (x1_ 1, xi) and from + CO to 0 in the interval (xk, CO) because 9 (x)-> co as xiand changes sign from - to + when passing through xi. Consequently, 9 (x)+A has a root in each of the intervals (xi_1, xi) and, besides, for A>0 one root in the interval (- co, x1), and for <0, one root in the interval (xk, + CO). Thus, cp (x)+ and, hence, f (x) [cp (x) + A]= (x)+ f ' (x) also has k roots different from x1, x2,_ x k for A00 or k - 1 roots different from xi, x2, ..., xk for A=0. Besides, of (x)+f ' (x) has x1, x2, xkas roots of multiplicity al-1, 0C2 1, .. -1. Thus, the total number of real roots (counting multiplicity) of + ak for WO and the polynomial of (x)+ f ' (x) is equal to oci+oc2 + + + + ak - 1 for A=0, that is, it is equal to the degree of the polynomial A f(x)+f' (x). 717. Let g (x)= a0(x+ Ai) (x +A2) ... (x+A), F0(x)= ao f (x), (x)= (x)= a0 f (x)+ao Alf' (x), F2 (X) = (X) + A2 (x)= a0 f(x)+ =F0 (x) + +ao(Ai+ X2) f (4 - a0A1X2 f" (x), etc. Then Fn(x)= F,,_ (x) + AnF,,_ 1(x)= anare coefficients of g. = ao f (x)+ ' (x)+ ... + a f (") (x) where ao, By virtue of Problem 715, all roots of all polynomials Fo, F1, Fn are real. 718. The polynomial ao xn+aimxn-l+ +m(m-1) ... (m- n + 1) an= In and all roots x'n are real. = [a0x'n +a2 (xm)' + . . . + an(xm)n.1 xn719. The polynomial anxn + nan_i x"- 1+ n (n - 1) an_ 2 Xn 2 + ... + ao n! has only real roots. Hence, all roots of a0n! xn + ain (n-1) . . . 2xn -1+ ... +nan_ lx+a are real. Applying once again the result of Problem 718, we find that all roots of the polynomial (io n! xn + ai n n (n- 1) ... 2xn- 1+ + ann! are real. It remains to divide + a2n1(n - 1) n (n-1) ... 3x" - 2 + by nl. n (n - 1) 720. All roots of the polynomial (1 +x)"= 1 + - x+ x2+ . . . 12 1 + x' are real. It remains to use the result of Problem 719. xn 2_ 721. The polynomial f (x)= ran 1 has a real root of 1. Furthermore, let F (x)= (x - 1) 1 (x)= nx" 1- (n + 1) xn + 1. Then F' (x)= =n (n+ 1) (x- 1) xn -1. For odd n, the polynomial F(x) has a unique minimum for x=1 and, consequently, has no roots except the double root x=1. For even n, the polynomial F(x) increases from - co to 1 for - co <x<O, decreases from 1 to 0 for 0 <x <1 and increases from 0 to co for 1 <x < co. Therefore, F(x) in this case has a unique root other than the root x=1. 722. The derivative of the polynomial that interests us is positive for all real values of x. Hence, the polynomial has only one real root.
252
723. Let a < b < c; f ( co) < 0; f (a)=Ba (6 a) + C2(c a)> ; f (c)= = A2(c a) B2(c 6) <0 ; f(+ co) > 0. Consequently, f has real roots in the intervals ( 0 o , a); (a, c); (c, + cc). 724. cp (a+ bi) 4 }P k(a a k bi) B+ a+biak B+ (a a k)2+ 62
k=1 n k=1
Im (cp (a+ bi)) = b
Aa k
k=1
(a ak)2+ 62
0 0 for b 0 0 because all terms under
the summation sign are positive. Hence, cp (a+ 6000 for b0 0. The same result may also be obtained from the fact that cp (x) varies from + co to co as x varies from ai to ai+i, cp (x) ranges from 0 to co for co <x< p (x) varies from + co to 0 for ;<x< co. It is assumed here that ai<az < ... <an.
n
725.
f ' (x)
f (x) Hence
k=1
v 1 x xk , where xkare roots of the polynomial f (x).
[f' (x))2- f (x) f " (x)=Ef (x))2

k=1
(X 1Xlc)B >
for all real values of x. 726. Let x, <x2 < <xnbe roots of the polynomial f (x), and let h. < <y2 < < y,nbe roots of the polynomial p (x). When the condition of the problem is fulfilled, m=n, n-1 or n+ 1. Without loss of generality, we can take it that x1<y1 <x2 <y2 < <y_,<x,, or xi < yi <x2 < y2 < <yn_,<xn <yn. We assume 100. Rewrite the equation as (x)= If m= n, then ,.1) (x) varies: from
fx
fix)
0 to co for co <x <y1, vanishing for x=x1; from + co to co for yk <x< yk +1, vanishing for x=xk +1; for yn <x< + oo. Here, ck, and 60are the leading coeffi0 cients of f (x) and cp (x), which we consider to be positive. Due to the continuity of 4) (x) in each of the intervals under consideration, the equation (x)= 1 ,has n real roots if 0 and n-1 real roots from + co to
ao
if =-T ) . Thus, the number of real roots of the equation f (x)+ tap (x) is equal to its degree. The case of m= n-1 is regarded in similar fashion. 727. The roots of f (x) and p (x) are necessarily all real since f (x) and p (x) are obtained from F (x) for A=1, 1.c.=0 and for 1.t.=1, X=0.
253
Suppose the roots of f (x) and cp (x) are not separable. Without loss of generality, we can take it that there are no roots of p(x) between two adjacent roots f (x) x1and x 2of the polynomial f (x). Then 4' (x)= is continuous for xl< cp () .<_.x.x 2and vanishes at the endpoints of this interval. By Rolle's theorem, there is a point x0inside (x1, x2) such that (x0)=0. Then 4) (x) (xo) has x0as a root of multiplicity Ic?,2. On the basis of the result of Problem 581, there are, on the circle I z xo I = p, if p is sufficiently small, at least four points at which Im (q) (z))=Im (4) (xo))=0. At least one of these points, z0, is nonreal. The number ti= (zo) is real. The polynomial F (x)= f (x)+ p. p (x) has a nonreal root, a contradiction, of the polynomial f' (x) divide the 728. The roots 1.< l2 < . < real axis into n intervals:
(-00, (c- 21 %-1), (Sry-1,
m).
By virtue of Rolle's theorem, in each of these intervals the polynomial f (x) has at most one root. Furthermore, the polynomial f ' (x) + V" (x) for any real has at most one root in each of the foregoing intervals. Hence, f (x) +V ' (x), by virtue of Rolle's theorem, has at most two roots (counting multiplicity) in each of the intervals. Separate all intervals into two classes. In the first put those which contain a root of f (x). In the second, those without any root of f (x). Consider the f ( x) function 4' (x)= . In the intervals of Class One, 4, (x) has one simple f' (x) root and therefore changes sign. In the intervals of Class Two, 41(x) does not change sign. In the intervals of Class One, 4)(x) +A has an odd number of roots (counting multiplicity). Hence, from the foregoing, 4, (x)+ X has only one simple root and no multiple roots in the interval of Class One. Therefore, 41(x) has no roots in the interval of Class One. Now consider the intervals of Class Two. Let &, be a point in some interval of Class Two in which the absolute value of 4) (x) reaches a minimum, and let A0 = 4)(W. For definiteness, we assume that 4, (x) is positive in this interval. Then the function 4) (x) has no roots in the interval of our interest when A<ao and has at least two roots when A> A0. By virtue of what has been said, the number of roots of 4'(x) X is exactly equal to two for A > X0and both roots are simple. Furthermore, 4) (4 A0 has &, as a multiple (double) root. Thus, 4, (x)-A has no multiple roots in the intervals of Class One and has only one multiple root for one value of A in each interval of Class Two. Furthermore, each root of the polynomial f' 200f (x) f" (x) is a multiple root for 4) (x) (71) since
i(P (x)] =
f' 2(x)- f (x) f " (x) [f' (x)]
Thus, the number of real roots of f'2(x) f (x) f" (x) is equal to the number of intervals of Class Two, which is obviously equal to the number of imaginary roots of f (x). 729. X (x)+ f 2 (x) has all real roots for arbitrary real constants A and p. (Problem 726). Hence, by virtue of Rolle's theorem, X f (x)+ f (x) has all real roots. From this it follows (Problem 727) that the roots of f 1(x) and f2] (x) are separable.
254
730. Suppose f (x) has no multiple roots and let x+ be the roots of f' (x). Consider the function 4) (x)= f(x) +
f' (x)
Y
<
. It is ob-
1 . ( x) n 1 = + >0 y >0 or if y < - n. Whence it folvious that lim y x->co x lows that 4, (x),+ co as x---> + 00 and 4) (x)--> - oo as x-* - co . Pesides 4)(x)-> - co as x-)-5i from the right and 4, (x),+. as x-- from the left. Thus, 4, (x) varies from - co to + CO in each of the intervals (- co, ...3 ( 3,_13 CO) remaining continuous inside these intervals. Consequently, 4) (x), and its numerator y f (x) + (x + A) f' (x) as well, has at least n distinct roots for y> 0 or y < - n. But the number of roots of yf (x) + +(x+A) f' (x) does not exceed n, because yf (x)+ (x+ A) f' (x) is a polynomial of degree n. If f (x) has multiple roots and x1, x2, ... xk are distinct roots of t from x1, x2, , xk. k-1 differen 23 f (x), then f' (x) has k-1 roots Reasoning in like fashion, we are satisfied of the existence of k roots of 4, (x). All of them, except - A if - A is among the roots of f (x), will be different from the roots of f (x). Besides these roots, y f (x) + (x + A) f'(x) will have as roots x1, x2, ..., xk with the sum of multiplicities n-k [if -A is not a soot of f(x)] or n-k+ 1 [if -A is a root of f(x)]. The total number of real roots of y f (x)+ (A + x) f' (x) counting multiplicities is again equal to n. (x +yd. Every yi is either greater 731. Let cp (x)= bk (x+ y1) (x + y2) than zero or less than n. It is _obvious that the coefficients of the polynomial
Fi (x)=ytf (x)+ xf ' (x) are ai (yi+i). The coefficients of the polynomial F2 (X)=y2 (x)+x F; (x) are ai (yt + i) (y2 + i) and so on, the coefficients of the polynomial Fk (X) = yk Fk (X) + XF1,_ 1 (x)
(yk +i), i=1, 2, ..., n. On the basis of the result of Problem 730, all roots of all polynomials Ft, F2, ...3 Fk are real. But aoco (0) + atcp (I) x+ . . . + a cp (n) xn = b k F 1, (x). 732. Suppose f (x)= (x) (x +A) where A is a real number and f1 (x) is a polynomial of degree n-1, all the roots of which are real. Suppose that for polynomials of degree n-1 the theorem is valid; on this assumption, prove it for polynomials of degree n. Let ft(x)=- b0+ btx+ +b,,_, xn-1,
are ai (y1+ i) (y3+
Then
-- ao+aix+ f(x)a, ----AN, a1 =Abi+ a2 =Ab2+ b1 , an_,=Thn_ an =
+a,,xn.
1)2 -2, b,
255
and
an+al yx+ a2 y (y-1) x1+ ... +an y (y-1) (y- n+ 1) x" =A [bo +bl yx+bo y (y- 1) x2 +... + b- a (Y-1) ... (y- n + 2) xn-l+xY [bo 2 + ... + (y-1) (y -2) + (y- 1) x+b,(y-1) (y-2) x (y- n + 1)xn-9=kp (x) + x[ycp (x)- x (x)]
where cp (x) is used to denote the polynomial
bo+biyx+b2Y (Y -1) x2 +
+b_iy (y-1) ...(y-n+2)
By hypothesis, all roots of the polynomial cp (x) are real. It remains to prove the following lemma. Lemma. If cp (x) is a polynomial of degree n- I having only real roots, then all roots of the polynomial (I) (x)= Xrp +yxcp-x2cr are real for y> n-1 and for arbitrary real X. Proof. Without loss of generality, we can take it that 0 is not a root of cp (x) for if cp=xlccpi, p1 (0)00, then (x)=xk (Xcpi+ (y- k) xcpi-x2cp',)= Xk4)1 and yi=y k exceeds the degree of cp1. Let x1, x2, ..., xn, be distinct roots of cp. The polynomial (1) has among its roots x1, x2, x with sum of multiplicities n-1 - m. Now consider w (x)= X+ yx It is obvious that urn
X-0.00
X2 9' (x)
cp
(x)
w (x)
=y-(n-1)> 0.
Hence, w (x)-- co as x -> - co and w (x)-+ co as x--->+ co. Besides, w (x)-> ->+ co as x->xi from the left and w (x)-o- co as x-xi from the right. For this reason w (x) has roots in each of the intervals (- oo, x1), (x1, x2), ..., x,), (x,, + co). The total number of real roots of cIJ (x), counting multiplicity, is equal to n- 1 - m + m+ 1 =n, that is, it is equal to the degree of 4'(x), which is what we set out to prove. 733. If all the roots of the polynomial ao + + ... + anx" are real, then all the roots of the polynomial aoxn+ a,xn -1+ ... + anare real. Furthermore, all the roots of the polynomials (y, n+ 1) x"+ al Yl (yi 1) aon (n-
n + 2) xn- 1+
and ao
+ an_ l y + a,,
1) ... (y1- n+1)+aln (Y1-1) ... (y1-n + 2) x + + a,,_ i y x"-1+ anxn al =[a0+ 1 x+ . . . + xn-1 a + 2) (Y0. -1) (n - a + 1 ) (Y 1n -n + a xnj yi (ri+ 1) . (y, -n+ I) 1 n+ 1) (yi - n+ 2) ... Yl (-1 ,
-
256
are real for y1> n-1. Setting yi-tz + 1=a >0, we find that all the roots of the polynomial
xn a0 + x+ x2 + ... + a.(a+1) a (a+n-1) a(a+1) are real. Using the result of Problem 732 a second time, we get the desired result. 734. 1. Suppose all the roots of f (x) are positive. Then the polynomial an + aiwx+ + anwnz xn cannot have negative roots. Suppose the theorem holds true for polynomials of degree n-1. Denote W=b0+b1 wx+ba w4 x2 +
+
a2
w(n-
Let 0 <x1<xa < <xn_ i where x1, x2, ..., xn_, are roots of p (x) and let xi > w -2. xi_i Further suppose that f (x)=(a .X) (bp+ biX+ ... + b._1xn -1). The coefficients of the polynomial f (x) are
a0 = Xb 0 , a1 =Abi 1)0 ,
a2 = Aba b1,
an _i=Abn -1- bn -2, n _1. an =-b
Consequently,
P(x)=c10-1- aiwx+aaw4 x2+ ... + anwO xn=A (bo - F b i wx

+ ... + bn _ l w(n -1) xn -1)-x (b0 w + bi w 4 x+ . . . +bn _ i we xn -1)
= Ay (x) - xwcp (xw2). The roots of the polynomials p (x) and x cp (xw2) are separated by virtue of the induction hypothesis. Thus, all the roots of the polynomial A p (x)+ +xw cp (xw2) that interests us are real. It remains to verify that the law of their distribution is the same as for the polynomial p (x). Denote by zi, z2, ..., zi, the roots of 4(x). It is easy to see that 0 <zi<xi<xi w-2 <z2 <x2 <x2 w-2 <zo < ... <zn_ i<xn_ i<xn_ i w-2 <zn. zi xi _ iw-2 Whence it follows that = w -2, which completes the proof. > zi_ i xi-1 X2 log 1 ) m 2. Consider cp. (x)= (I
m Iv
For m sufficiently large, the roots of the polynomial Pm-(x), equal to V m , do not lie in the interval ry (0, n). Consequently (Problem 731), 1 log w all the roots of the polynomial an op. (0)+ ai yn, (1) x + ... + an yn, (n) xi' are real. But lim Pm (x)=wx'. Hence, by virtue of the continuity of the roots m->co as functions of the coefficients, all the roots of an+ai wx+ ... + anwo xn are real.
257
x,, the roots of the polynomial f (x)-=- ao + aix+ 735. Denote by x1, x2, + an xn. Without loss of generality, they may be considered positive. Furthermore, let cP (x)= ao cos cp + alcos (cp+ 0) x + LP (4= aosin cp+alsin (cp + 0) x + Then
p (x)+ (x)= (ccs + i sin cp)
+a, cos (p + nO) xn, + ansin (cp + nO) xn.
11 (ax x,), ,=.

(a' x x,)
(i) (x) i4 (x) = (cos p i sin cp) an where Consequently, 413. (x)=
p
a -= cos 0+ i sin 0, a'= cos 0 i sin 0.
(x)+
(x)
i= Let x= (DP be a root of the polynomial cp (x). Here, P=I x I; p=cos + i sin A. Then I (x) I = 1 and, hence, n
cp (x)i4) (x)
cos cp + isin p cos cp i sin p
11
axxi x xi
i=I
Pa' p xi
Papx1
=I
,
but (-,a(3 xi pa' px,
12
(Pap xi) (Pa'
P' xi)
pxi (a a') ((3' p) =1 +4oxi ' sin (i) sin pa' (3 xi 2 i pa' p xi 2
(Pa.' (l xi)(Pap' xi) =1+
We disregard the uninteresting case of sin 0=0. Papxi If sin A00, then all 2 are simultaneously greater than unity pa' p xi or simultaneously less than unity and their product cannot be equal to 1. Hence, sin A=0, which means x is real. 736. Let x1, x2, ..., x be the roots of the polynomial
f (x)=ao+ ibo + (ai + ibi)x +
+ (a,,+ ib,,)xn=cp (x)+ i (x).
The imaginary parts of these roots are positive. Let us consider the polyno4 x,, ' , which mial f(x)=p (x) (x). Its roots will obviously be are conjugate to x1, x 2, ..., xn . Then
n
(I) (x)_
(x)+i4) (x)
(x)i4) (x) 1 i=i
XXi
an + ib a ib
9. 1215
258
If xois a root of op (x), then

I CD (X0)1 =
X 0Xi
1.
But
xo xi 12
i=i
(X0 Xi) (XP X;) 1
X0 X;
(Xi
(X0 XP)
X0 X;
(X0 X;)
Xi)
I X0 Xi 12
=1
4 Im (xo) Im (xi)
I X0 12
Whence, if Im (x0) > 0, then

X o
xo xi ease and without computations.) Thus, the equation I (1) (x0) I =1 is only possible for a real xoand therefore all roots of cp (x) are real. Next, consider the polynomial (a pi) [cp (x)+ i4) (x)] = cap (x)+ p4 (x)+i [0,4) (x) pcp (x)].
<1 for all i; if Im (x0)< 0, then xo xi >1 for all i. (The same thing can be obtained geometrically with
xo xi
Its roots do not differ from the roots of the original polynomial and, hence, its real part ay (x)-+ t) (x) only has real roots for arbitrary real a and p. But in this case, the roots cp (x) and 1) (x) can be separated (Problem 727). 737. Let x1 x2, x be the roots of cp (x); yl, y2, ynthe roots of c,I) (x). Without loss of generality, we can assume that the leading coefficients of cp and 4) are positive and
,
> y1>x2 > y2 > > (yn may be absent). Decompose (x)into partial fractions (x)
(x) A + cp (x)
> > Yn
k=1
E Ak xxk
Ak
(xk)
It is easy to see that all Ak >O. Set x=a+ bi and find the imaginary part of (cp (x)+ (x)) cp (x) Im (x) cp (x) '
:, 4 (x)
i) =
1+1m
(E
k=1
Ak
a+bixk
= 1b
E
k=1
Ak (a x k)2+ 6'
259
(x) i I <0 and, hence, (p (x)+ (x)00. Thus, If b>0, then Im cp ( x) in the case at hand, all the roots of cp (x)+ (x) lie in the lower half-plane. The other cases of location of roots are considered in similar fashion.
738.
f'(x) _ \-,
f (x) Let x= a bi, b > 0. Then
1 L x X k ' X k are roots of f
(X).
k =1
Im Consequently,
f ' (a bi)\ _ v f (a bi) L-1
b+Im (xk) >0.

I )c xkl 2
k =1
f' (a bi)0 O. 739. Let the half-plane be represented by the inequality r cos (0 cp) > p, where x= r (cos (p + i sin cp). Put x= (x' + pi) (sin 0 i cos 0). Then x'= pi+ x (sin 0 + i cos 0)= r sin (0 cp)+ i [r cos (0 cp) p]. Whence it follows that if x lies in the given half-plane, then x' lies in the halfplane Im (x')>0, and conversely. The roots of the polynomial f [(x' +pi) (sin 0 i cos 0)] are thus located in the upper half-plane. On the basis of Problem 738, the roots of its derivative, equal to [sin 0 i cos 0] f' [(x' +pi) (sin 0 i cos 0)], also lie in the upper half-plane. Thus, the roots of the polynomial f' (x) lie in the given half-plane. 740. This follows immediately from the result of Problem 739. 741. The equation splits into two : f' (x) f' (x) 1 + =0 and f (x) f (x) ki Decomposition into partial fractions yields 1 =0. ki
k=1
1=0 x_ xk+ki
xk are the roots of f (x), which, by hypothesis, are real. Let x= a+ bi. Then Im E
k=1
XXk
) =I b I
k =1
n (a x k)2 + < b I 1
For the roots of each of the equations it must be true that -k < b
whence ,
I bl<kn.
9*
260
742. All the roots of f' (x) are obviously real. Denote them by Sl, 2i Next, denote by yl, Y2, the roots of the polynomial f (x)-b, by x1, x2, ..., xnthe roots of the polynomial f (x) -a. Then
, Yi< 3.<Y2<2<
xl < 1 < x2 <
< xn _ i< c- 3.<xn.
From these inequalities it follows that intervals bounded by the points
xi, yido not overlap since they lie in the nonoverlapping intervals
(- 00, W, + CO.
The polynomial f (x) takes on the values a and b at the endpoints of each of these intervals and passes through all intermediate values inside the interval. Consequently, f (x) - A vanishes n times on the real axis, which completes the proof. 743. If the real parts of the roots of the polynomial f (x)=xn+aixn-l+ + anhave like signs, then the imaginary parts of the roots of the poly+ nomial
inf
ia3x" -3+ (ix)xn+iapcn la2xn -2 -
also have like signs, and conversely. For this, by virtue of the result of Problems 736, 737, it is necessary and sufficient that the roots of the polynomials xi' - a2xn-2+a4x"-4- ... and a1xn -1-a2xn -3 aocn -5- ... be real and separable. 744. It is necessary that a> 0 and that the roots of the polynomials x3bx and ax2 - c be real and separable. For this, the necessary and sufficient condition is 0 < <b or c > 0, ab- c > 0.
Thus, for negativity of the real parts of all roots of the equation
x3-Fax2 H-bx+c=0
it is necessary and sufficient that the inequalities a>0, c>0, ab - c>0 be fulfilled. 745. a >0, c>0, d>0, abc- c2-a2d>0. . It is easy to see that if x I< I, then the real part I -y of y is negative, and conversely. Consequently, for all roots x1, x2, x 3of the equation f (x)=0 to be less than 1 in absolute value, it is necessary and sufficient that all the roots of 1 +y the equation f ()=0 have negative real parts. This equation is of 746. Set
x- 1 -1-y
1y
the form y3 (1 -a+b-c)+y2 (3-a-b+3c)+y (3+a-b -3c)+(1 +a+b+c)=0. Besides, it is easy to see the necessity of the condition 1 - a+ b - c=(1 + xi) (1 +x2) (1 +x3)>0.
261
On the basis of the results of Problem 744, we get the necessary and sufficient conditions:
1a+bc>0, 1+a+b+ c>0, 3 ab+3c>0, 1 b-Fac c2>0. 747. f (x) (1 x)= an + ( a,,_ 1 an ) x +(a,,_ 2 - an_ i) x2+ ... +(ao ai)x" aoxn+ '
Let x I -= p >1. Then
f (x) (I x) I aopn+
I an+ (an _
a,,) x
ao Pn + 1P"
+(ao al) xn I
(an+ an-3. an
+ aoai)= ao (Pn+ 1 p9> O.
Consequently, f (x) 0 0 for 1 x I> 1. 748. 0.6618. 749. 2. 094551. 750. (a) 3.3876, 0.5136, 2.8741; (b) 2.8931; (c) 3.9489, 0.2172, 1.1660; (d) 3.1149, 0.7459, 0.8608. 751. The problem reduces to computing the root of the equation x3-3x+ +1=0 contained in the interval (0, 1). Answer: x=0.347 (to within 0.001). 752. 2.4908. 753. (a) 1.7320, (b) 0.7321, (c) 0.6180, (d) 0.2679, (e) 3. 1623, (f) 1.2361, (g) 2.3028, (h) 3.6457, (i) 1.6180. 754. (a) 1.0953, 0.2624, 1.4773, 2.3556; (b) 0.8270, 0.3383, 1.2090, 2.9563; (c) 1.4689, 0.1168; (d) 8.0060, 1.2855, 0.1960, 1.4875; (e) 1.5357, 0.1537; (f) 3.3322, 1.0947, 0.6002, 1.8268; (g) 0.4910, 1.4910, (h) 2.1462, 0.6821, 1.3178, 4.1463.
CHAPTER 6 SYMMETRIC FUNCTIONS

755. The following is a detailed solution of Example (f):
F(x1, x2, x3) = (xi+ xi) (xi+ xg) (x2+ x3)

The leading term of the polynomial F is xi 4. Write out the exponents in the leading terms of the polynomials that will remain after a successive elimination of the leading terms due to subtracting appropriate combinations of the elementary symmetric polynomials. The exponents are: (4, 2, 0), (4, 1, 1), (3, 3, 0), (3, 2, 1) and (2, 2, 2).
262
Hence, F=f,2 f+Af i2f 3+Bf23+C f, f 3+ Df32where A, B, C, D are numerical coefficients. We determine them, specifying particular values for x,,
X1, x3.
X2
X2
x3
f1 2 0 3 1
fa
f3
1 2 1 1
1 1 2 1
0 1 2 1
1 3
0
1
0 2 4 1
2 50 200 8
We have the following system of equations for determining A, B, C, D: 2=4+B, 50= 27B+4D, 200= 108 A + 16 D, 8=1 AB+C-FD whence B= 2, D= 1, A= 2, C=4. Thus, (x1+4) (4+ 4) (4+ .4)=fM 24f The answers for the other examples are: (a) P .3fif2; (b) fife -3 f 3; (C) f 4.fi2f2+8.fif 3 ; (d) (h) ft 2fif 3
3 f1 f1+6N2f3+3.af 3
2/1+ 4f1 f2 f f32.
(e) fif2 f3; (g) 2f 9.A.f2+ 27/3 ;
+ 18fif2 f3-2742.
756. (a) fif2 f 3 fPf4 A; (b) Af4+fi 4f2f4;
(c)
(c)
f3 f 3 4f1f2 + 8 3 f3 ;
757. (a) fl -2f2; (b) f 3 3fi f2

fif 3 4f4;
(d) a 2fif 3+ 2f4;

4fi2 f3+ 2f1+ 4fif 3
4f4;
2 fif3-2f2+4f4; (f) (e) f?./.

(i) f1f1-2fif f2 f 3 + 5 f1f4
) frf2 3 f1
(g) f2f3-3.fif4+5f3; (h) fff 3 2f2f 3 fif4+5.1.3;

5 f5 ;
4-5f3; fPf3+5.fa f 3 +.11.1.
(k) f2-5f;f2+5f1n+5.1?f3-5f2f3-5.11.1.4+5f3;
263
2f2 f4+ ?fa fa 2f6 ; (n) fa2 f4 2 fa + 6A ; f2 f4 (0) f2 13 3flf4 3fR+ 4f2 f4 + 7fif2 12f6; (P) f; f2 fa f 3+ 3112f4+ 3f: 3f 2 f4 3fi 3f6 ; (1)4f11f5 +9f6; (m)
- 3
(q) R/3
f 3 fif4
3f1+ 2f2f4+
6fo;
(r) fpn 2fif 3
+ 2f2f4 6fi fa + 6f6; (s) fif2 4/1 2f flfa + 2f;+ 7fif2 f 3 f1f4 311 MaMs+ 6f6;
6frf 3
2f; 12f1f2 f3-6f12f4
2f 3 VI fa fa +
(t) ff 6f1f2+ 9Pf
+3.fl+ 6f2 f4+ 6.fif
6fo.
758. (a) nf 2 8f2; (b) + 4fin 2 .1T-3A+...+( 2)n h. f2 -8 759. (a) (n 1) f-T 2nfa ; (b) (n 1) f 3 (n 2) f2+ 3(n 4) f 3 ; (c) (n-1) ft 4nf?.f2+ 2 (n + 6) f 4 (n 3)11 f 3 4nf 4; (d) 3 (n -12 1) (n 2) (3n 1) (n 2) f2.
2fk 3 fk + 3+
760. )7 2fk a fk + a + 2fk-2fk + 2 761. (n
...
2
I)! E
i=1
f 2 2 (n-2)! [n
E
i=1
q(2
a i) if2
i=1 = (n 1)! S2 S2+4 (n 2)! F2f2
where
S2 =
E
f3
aF; s2= E xF;
i=1
F2 =
i<k
ai ak; f2=
E
i<k
xixk.
762. (a)
f2 3f 3
, (b)
(fif2
f 3 2f1)
3.f (c) f2 + f fi 2fa
f l f2f 3 fa fa + 9.f?
(b) f 2f _ 2 fn
f 3 6 f1 f2 fa+ 6f3+ 2f12 f4
f L./2/.3 ftfaf:
763. (a)
fa + 2f4
(b)
f4
fn 1 fn
.
764. (a)
fn
; (c)
fn 1 nfn fn
(d)
f?g__ 2f2 _ 1 fn 2f.+ 4f2 fn 2 fn
(e)
fn 1 fn 1 fn fn
(f) f2 fn 1 (n 1)
fn
fn
264
765. 4. 766. 35. 767. 16. 768. (a) 3; (b) 2p3 -3q2 ; (c) 133 (xix2 x3 (d) 94; (e) 2p 3q 1+ p q p); = 2p2 4p 4pq+3q2+6q (f) (1 +p q)2 2b. Hence, 1/ a2 2b V 2
769. Let x2=4+.4. Then 24=4+4+4= or
V a2 2b i among the roots of the given equation. For this, it is is 2 necessary and sufficient that the following condition be fulfilled: a 4 (a2 -26)=2 (a32ab +2c)2. 770. a= ..x1x2 X3, ab c= (xi+ x 2) (x1+ x3) (x2+ x 3),
C= x1 x2 X3.
If all roots are real and negative, then a>0, b>0, c>0. If one root x, is real and x 2 and x3are complex conjugate roots with negative real part, then xy X3 <0, Xy X3 > 0, (x1 +x2) (Xi + X3) > 0 and, hence, also a> 0, b> 0 and c> 0. The necessity of the conditions is proved. Now assume that a> 0, b> 0, c> 0. If x, is real and x2 and x 3are complex conjugates, then x2 x3> 0, (xi+ x2) (xi+ x3) > 0 and from c> 0, b >0 it follows that xl <0, 2Re (x2)=x2 +x3 <0. Now if x1, x 2, x3are real, then from c> 0 it follows that one root, x1, is negative, and the other two are of the same sign. If x2 >0, x 3 > 0 , it follows that
- X1- Xy > X3> 0, - X1- X3 > Xy > 0
and then (xl -Fx2) (xl+x,) (x2 +x3)<0, which contradicts the hypothesis. Hence, x2 <0, x3 <0. An alternative solution is given in Problem 744. 771. s= 1 -,
v a (4ab a' bc), R
V a (4ab as bc)
772. a (4ab a38c)= 4c2. 773. (a) 774. (a) 25 , (b) 35 ,679 (c) 1 a0 + 18a, a, a2 a, 274, 4; a; a, al ao.
al 4a? a,
(b) ai a, ao ; (c) al a2 9; (d)
265
775. It is sufficient to give, the proof for the elementary symmetric polynomials. Let ykbe an elementary symmetric polynomial of x 2, x3, ..., xn of degree k; let fk be an elementary symmetric polynomial of x1, x2, ..., xn. It is obvious that cpk=fk xi Pk-1, whence it follows that
(
Pk =fk
fk-1.
(_
cpk= ak+ ak_i
776. xi + x2=f1x3, (f1 Xi) (f1 xi)(f1 x3)= P fi+fif2f2=fifzf,, 2x, x2 x3= 3x,f,. (3x1fi) (3x2 fi) ( 3x3 A)= 27f 3 9ft fi +
xl+ x1 x2+ =f? f2

xy X2 X3 =fl
x3,
777. E axi = kit k)rfk 1

afk
=1
778. Let F (xi, x 2, ..., x

n
(I) (f1, f2, 4). Then 072- + . . . +f_i fn (n 1) f,
E =
i
OF
oxi
=n
all)
Oa)
779. Let co
F (xi+ a, x2+ a, ..., xn+ a). Then p (a) = dF (xi+ a, x2+ a, . , xn + a)
i=1
n
axi
Since cp (a) is not dependent on a, then cp' (a) is identically zero, whence it folthat i= 1
aF =
Conversely, if
i =1
aF(xi, x2, ..., x) is iden axi
aF (x,a, x+a) =0 whence it follows tically zero, then (f,' (a)= axi 1=1 that cp (a) does not depend on a and 9 (a) =cp (0), that is, xn). F (xi+ a, x2+ a, ..., x,,+ a)= F (xi, x 2,
By virtue of the preceding problem, the condition lent to the condition
=0 is equivaE aF ax,
t
- aft
+(n 1) fi af2 +
ae afn =o.
266
780. Let F(x1, x3, ..., x,) be a homogeneous symmetric polynomial of degree two. Then its expression in terms of the elementary symmetric polynomials is of the form (1)=Af,2+Bf2. By virtue of the result of Problem 779, it must be true that n 2Af1+ (n-1) Bf1=0, whence A= (n 1) a, B= 2na and
F(x1, x 2, . . . , x,,)= a [(n 1) f? -2nf21=
E (xi-x02.
i<k
781. The expression of a homogeneous symmetric polynomial of degree three in terms of the elementary polynomials is of the form Aft+Bf1 f2+ + Cf B. By virtue of the result of Problem 779, it must be true that 3Anff+ + nBf2+ (n 1) Xi+ (n 2) Cf2=0 whence + 3n2f 3]. F(x1, x 2, . . , x 3) = [(n 1) (n 2)p - 3n (n 2) 782. (n 2) f?f 2 (n 1) fif 3 4 (n 2)f: + ( 1 On 12) f,..f2 f 3 4 (n 1)42 1'4 9nf2 + 8nf2 f4. 783. We can take
pk=sk (x,
-A, x -
-L I )
Each function Pkhas the required property. Furthermore, if F x2, ..., xd= =F (x1+ a, x2+ a, ..., xn+ a) and F(x1, x2, ..., xd= 4 (fi, f2,f 5), then
F(x1, x2, xn) = (I) cp2, 93, (1),,) 784. (a) 4cpg 2791 (b) 184 785. (a) 893, (b) 494(4+169194 -2791+1 4 4 92 ,494 -12892 91+ 2569i 786. s2 =f1-2f3 ;
s3=f?-3 fif2+ 3f 3 ;
s4=fl 41. 12f2+ + 4f1 f 3 4f4; s5=f1-5f?f3 + 5fif + 5f,21.3 5f2 f3 s6= 6.1 .1f2+9f 2 f1+ 6fi'f 3 5 ./. 1f4+ 5 .f5;
6f12f4 + 6 f2 f4 6 f1 f5 6f6
2f1 1 2 fif2 f3
787.
2f2=4 s2; 6f 3 = 3,5'1 S2 + 253 ; 24f4= st 6s1 s2 + 8s1 s3+ 352 654 ;
I 20 = St 1 OS? S2 + 204 s3+
15s1s2-2052 S3 30.si 5 4 + 2455; 720f6 =4154 53 + 40s1 s3+ 45 51 .s2 120 si s 2s 3
154-904 s4 + 404+ 9053 s 4+ 144 51 s3 12056790. 788. s5=859. 789. s3 = 13. s10= 621. 791. si= 1, s2 =s3 = ... = sn=0. 792. This is readily proved by mathematical induction by means of the relation
ask+ bsk _ i+ csk _ 2 =0

where sk=x+4:.
267
793. s6 s1=5(fff2) (f3 f1f2);
s3 s1 =3 ( f 3
f2).
794. S5= 5f2f3; s3=3f3; S2 = 2f2. 795. s,= 7f2 f5 ; s2= 2f2 ; s5 =5./3.
796. e -a=0.
797. x" xn-i+ 798. xn+ P
' (a)
a -1-
a2 2 xn-2
. .
xE
a _ n! =0
P n (6') 0, where P1., P2, ...3 Pri
xn-1+P2 a) Xn- 2 . 2(!
n!
are Hermite polynomials: Pk(X) = ( 1)k e 2 dxk Hermite polynomial Pn+1(x). Solution. Let the desired equation have the form
xn+a1 xn-1+a2 x"-2+...+a,,=0
d ke
2
a is a root of the
x2
By virtue of Newton's formulas a1 =a, 2a2 =aa1-1, 3a3 =aa2 al, kak
a k- 25
nan =aan_, an _ 2,
= ccan an_1.
From these relations it follows that ak is a polynomial of degree k in oc. Set k!ak=Pk (a). Then, taking P0=1, we get P1=a and PkCC Pk_ 1(k I) Pk _ 1=0, aPn+nPn _ 1=0. The first relations show that Pk is a Hermite polynomial in a (see Problem 707). The latter yields Pn4.1 (a)=0. 1 r, (sz 799. 2 k s3k)-
800. (x+xi)k = v cm k sk me!, E =1 m=0 E E ( xi+x,)k= I

i=1j=l In=0 n n Cl nSk-m Sm,
E
i <j
(E
m=0
cr sk_msni_2k sk).
268
PART III. ANSWERS AND SOLUTIONS 2k V '
1 801. E (x,_xir k= -2
2.1 0 21
1 Y"Sm Szk m
m=0
802. Multiply the second column by s1, the third by s2, ..., the kth by (-1)k -1 s k_ iand add to the first. By Newton's formulas, we get 0 0 f1 1 0 f, 1 . . . 2 f2
(k 1) fk -1 fk -2
kfk A-2
f1
1'11
0 0
0
1
f,
fk
Sk fk-2
0
1
fk
... 0
. . .
0
Sk.
1)
fk-2
803. Multiply the second column by f1, the third by f2, ..., the kth by (-1)k -'j k _ 1and add the results to the first column. Newton's formulas yield the desired result.
805.
804. n! (x" f ixn-if2xn- 2+ cp (n) 1n P ( 71 n) k 71
(_ I )nfn).
where d is the greatest common divisor of m
and n. 806. By virtue of the result of Problems 117, 119, it suffices to consider the case n=p1 p2 pk , where pi, P2 ... pk are distinct odd primes. In this (-1)il-1if n is divisible by 3, and s3 =( 1)k case, s1=s2 =s4= (-1)k; s if n is not divisible by 3. tations eompu by Newton's formulas yield:
f2=
1 ( 1)k 2 _ i)k -1_ I 2

1)k
f3f3 =
if n is divisible by 3, 1 if n is not divisible by 3, if n is divisible by 3,

if n is not divisible by 3.
f4
fi = 807. s1=s2 =s3=
2 (_o k-i+i 2
= sn= a. Hence, for
kfk = afk _, afk _ 2+ + 1)k 1fi, (k 1) fk_1=afk-2+ +(-1)k-2fi,
whence
kfk= (a k + 1) fk -1, fk= a
+1
ik-i
269
Obviously, f1= a; therefore, f2= a(a-1) a(a-1) (a-k+1) 1.2 ' '' fk 2... k and so x1, x2, ..., xnare roots of the equation
a ( - I) xn-2 . .. ( _ 1),, a (a-1) . .. (a- n+1) =0, a x" - - x"-1+ 1a .2 n! 1 sn+i=a a (1- a) (2-a) . . . (n - a) n!
808. (x - a) (x - b) [x" + (a + b) x"-1+ . . . + (a" + an-1b+ . . . -F bn)]= a 2 xn-i + x _ b (an an-1 b+ = (.7C a) [x"+1 +b")] bn+i )x+ab(an+an-l b+...+b" ). _ xn+2_(an+i +an b+
The power sums c2, ..., anfor the new polynomial are obviously equal to zero. But ak =sk +ak + . Hence, sk = -(ak+bk) for 1 809. sk = -ak bk for odd k,
c k
Sk =
k \2
2 b 2 ) for
even
810. (a) (x+ a) (x2+ax+ b)- c=0; (b) x (x- a2 + 3b)2 -(a2b2-4 a' c - 4b3+ 18abc - 27c2) = 0 ;
aa c =0 ; (c) x3+ (3b - a2) x2+ b(3b - a2) x + (d) X2 (X a2 +3b)+ (aa b2-4a3c-4b3+ 18abc -27 0=0; (e) x3-(a2-2b) x2+ (b2-2ac) x- c2=0; (f) x3+(a3-3ab + 3c) x2+ (1)3- 3abc + 3c2)x+ c3 = O. 811. y2+(2a3-9ab +27 c) y+(a2-3b)2=0. b3+ a3c -6abc +9c2 ab -3c _ 0. 812. y2 y+ C2 3 -5abc + 6c2 4 y5+ b 1 c2 b3-5abc +6c 2 a2 b2 -2b3-2a2b + 6abc -7 c2 + c2 c2 814. (a) y' - by2+ (ac - 4d) y -(a2d+ c2-4bd)=0 813. y6 ab -3c
y2
ab -3c
y I 1 = 0.
--
(Ferrari's resolvent),
(b) y3- (3a2- 8b) y2+(3a4 -16a2b +1662 + 16ac - 64d) y- (a' -4ab+8c)2=0
(Euler's resolvent),
(c) y6 - by5+(ac- d) y4- (a2d+ c2-2bd) y3+ d (ac- d) y2- bd2y +d3=0; (d) y6 + 3ay6 +(3a2 +2b) y4 + (a3+4ab) y3+ (2a2b + + ac-4d)y2 +(ab2+ a2c - 4 ad) y +(abc- a2d - c2)=0.
270
815. x=
a+ -Va2 46+ 4y, Va2 46 +4y2 142-46 +4y2
4 The signs of the square roots must be taken so that their product is equal to
a3+4ab-8c.
816.
3 X 33
1/4a+e2 V b2 -64a3 11 4a+1/ b2 64as +-I/4a +E Vb2 64a3 2

e
1 i V3 = 2 + 2
The signs of the square roots are taken so that their product is equal to b. 817. (y+a)4 (y2+6ay +25a9+ 3125b4y = O. Solution. The roots of the desired equation are:
x2+x 2 x3 +X3 X4+X4 X5+X5 Xi) (X4 X3+X3 X5+X5 X2+X2 X4+X4 X1);
Y2 = (X1 X3 +X3 X2 +X2X5 +X5 X5+X4 Xi) (X4 X2 +X2 X4 +X4 X5 +X5 X5 +X5 X1); Y3 =(X6 X2+X2 X4+X4 X3+X3 Xl+Xi X5) (X5 X4+X4 Xi+Xl X5+X2 X3+X3 X5); Y4 =(X2 X1+X1 X5 +X3 X5 +X5 X4 +X4 X2) (X2 Xs +X5X4+X4 X1+X1 X5 +X5 X2); Y5 =(X5 X3+X3 X2+X2 X4 +X4 X4+X4 X5) (X5 X2+X2 X1+X4 X3+X3 X5+X4 X5);
Y6= (X2 Xi+XI X4X4 X3 -FX3X5+X5 X2) (X2 X4X4 X5X5 Xi+Xi X3+X3 X2).
The sought-for equation is obviously of the form y6 + c1ay6+ c2a2y4+ c3a3y3+ c4a4y2+(c6a2+ c6b4) y+(c7a6+ c8ab4)=0, where c1, c2, c3are absolute constants. To determine them, put a= 1 b=0, and a=0, b= I. We get
a
1 0
X1
x2
x3
x4
x5 0 1 0
Y2
Y3
Y4
Y5
Y6
0 1
1 1
,2
3-4i 5
3+4i 1
E3
Sc3 5e2 5e
In the first case, the desired equation is of the form (y 1)4 (y2 6y +25)=0. In the second case, y6 + 3125y= 0, whence we determine all the coefficients, except c3. It is easy to verify that c5=0. To do this, we can, say, take a= 5, b=4. In this case, x, =x2 = 1, and the remaining roots satisfy the equation x3 + 2x2 + 3x +4 =0 and all the necessary computations are performed with ease.
271
818. Let f (x)=(x xi) (x x2) ... (x xn), where xi, x2, ..., xn are independent variables. Also, let xk (x)=f (x) qk (x)+rk (x) and rk (x)=cki + ck2 x+ . . . ckn xn -1. The coefficients cks are obviously some polynomials in x1, x2, ..., Furthermore,
C11 C21 Cny C12
C22

Cyn
C2Il
1
Xy xn--1
1
X2 xn-1

Xn xn-1
en2
Cnn
r1 (x1) r2 (xi)
r1 (xi)
r1(Xn)
r2 (xn)
r2 (x2) -
rn (x1) rn (xi)
rn (xn)
cP (xi) xi eP (X1) xn-1cp (x
p (xi)
X2 eP (x2)
5 (xn)
Xn CP (Xn)
..
cp ( x2) .
1
X7, -1 (I) (Xn)

1 Xy . 1 Xn X n -1 I
=(13(xi) cP (x2) ... cP whence it follows that

C12
C21
..
(xn)
Xy
,c7
Cyn C2n Cnn
=cP
.7e4-1
C22
Cu2
(xi) (f) (xi)
cp
(xn) =R
cp)
eny
The last equation is an identity between the polynomials in the independent variables xi, x 2, ...,xnand therefore remains true for all particular values of these variables. 819. First of all, satisfy yourself that all polynomials ti) k (x) are of degree n-1. Introduce the following notations: fk (x) = an xk ak _1,
fk(x)= ak X nk + + an,
Pk (x) = bo x k 1+
(15k (x)=bk xnk+
bk-1,
+ bn.
Then
(x)= Xn k +1A (x) +.7k (x),

cP (X)=Xnk+1cPk (x) -F cf,k (X), t (X)= f k (x)[Xn k +1 cPk (x)+(P k (X)) cpk (x) [xn k +1-fk (X) +7 k (X)] = fk (X) ci) k fk (X)=(ao bk b0 ak) xn 1+ .
272
Suppose (1)k (X)=Ckl+ Ck2X+ ... H-CknX11-1and let x1, x2, ..., xn be the roots of the polynomial f (x). Then c11
C21
c12
Cin 1
11
1
X2
1 Xn
C22 C2n 1 . X1
Cn1 Cn2 . Cnn
x? -1
LPL Lli0
...
(X1) LP1
(X2)
I
4'1
(Xn)
(X1)
4) 2 (Xi)
LI) 2 (xn)
n (X1) `in (Xi)
4 n (Xn)
(xn)
(xn) p (xn)
fi (X1) p (xi) fi (Xi) (xi) fi (Xn)

= ; f2 (xi) (xi) f2 (Xi) co (x2)
fn (xi)
(X1) fn (x2) p (Xi)
fn (xn) (xn) f2 (xn) fn (xn) 1 1

X2
If (x2) A. (x2) f1 (xn)
-=-P (xi) (I) (x2)
ep (xn) ' no 0 a0
12 (xi) f2 (Xi) fn (xi) fn (x2) 1 ... 0 1 ... 0 xl
=9 (x1) p (x2)
(xn)
a_, an_2 a 0
x" -1
. . . x'n2 I
1
X2
1
= (Xi) p (x2) p (xn) cp
X1
Xn
Xn -1 xn 1 2 " Xn n 1 1
whence it follows that C11

Cul C12 C22
Cin C2n =0,02 p
. . ..... . .
(x.t) co
(x2) cp
(Xn) =R
5 0 )
Cn1
Cn2 Cnn
820. The polynomials X k have degree not above n-1. This is obvious for 1 and for k > nm it follows from the fact that X k are Bezout polynomials t.Pk _ n+m for f (x) and xn m cp(x). Let xk (x)= cki + c k2x + . + cknxn 1 and 1 1
A= Then C11 C12
Cln Xl (X1) Xl (X0) Z1 (X0) !
x,
1
X2
x,,
.
xn n -1
C21 Cyy C2n I
A = X8 (xi) Z2 (x2) 7,2 (xn)

I Xn (x1)
\
Cn 1
Cn2
Cnn
Xn (X2)
Xn (xn) i
=-- (P (xi) cP (x2) ep (xn)
273
x n-
m-i
xn
- in- I
x7 'n fo (xi)
fo (x2) 'n fm _ i ( xn )
x7 -m
1 -rn fm _ i (x2) ... (x1) x'2
=W (Xi) (P (X2)
cp (x) A
ao a,
ao
(11,1_2 am-2
cp (x1)
an (x,) A
whence immediately follows the desired result. 821. (a) -7, (b) 243, (c) 0, (d) -59, (e) 4854, (f) (bu (+2-62a0)2-(bo a1-b1ad (b1 a2-b2 a1). 822. (a) For A=3 and A= -1;
(b) A=1, x= -2+1/2
V41/ 2 -2
2
A=
-2-1/2 +il/ 41/2 +2 2 X= 1/TD,
(c) X=
823. (a) y6 -4y4 +3y2 - 12y + 12=0, (b) 5y0-7y4 +6y3- 2y2-y- 1 =0, (c) y3+ 4y2-y- 4 =O. 824. (a) x4=1, x2 =2, x3=0, x4 = -2, Y1=2, Y2=3, Y3 = - 1, y4 =1. 2=3, x3=2, x4 =2, (b) x1=0, x Y1=1, Y2=0, Y3=2, Y4=-1.
(c) X1=X2= 1; x3= -1; x4 =2, ; Y3 = 1; Y4= 2. Y1=Y2 = -1
(d) x1=0, x2 =0, x2 =1, x4=1, x4,4 =2, Y1=1, y2=3, Y3= 2, Y4=3, Y5,6 = 1 V 2 . (e) x1=0, x2=0, x3=2, x4=x6=2, x6 = -4, y1= 2, Y2= -2, Y2=0, Y4=Y5= 2, y6 =2, x7 =4, x8 = -6, xs= -213, y7=6, y8=4, y2 =4/3.
10. 1215
274 825. ag ar,iz I
826. Let f (x)= (x xl) (x x 2) . . (x x,,);

(x)= bo Xk
+bk; cp2 (x)= co x'n +
+ cm.
Then
/2
R (f, ip1 .PO= aT -Fk

1=1
pit (xi) cps (xi)
qa
R(X,,,xn'l)=
bn 1 1 pi (xi)] [ o
i=i
11 cp. (xi)]= R(f, P.1) .R(f; PO.
827. Only the case n> 2 is of interest. Denote by d the greatest common ... are primitive nth roots of unity; 7)1, divisor of in and n;
n =n1. Then, are primitive roots of unity of degree
fl
_1) = n (1
(n) p (n) = in (1 701(P (14) [Xn,Wi g) (n')

If in is divisible by n, then R (X,, xm 1)=0. But if m is not divisible by n, then niO 1, and, by virtue of Problem 123, X.1 (1)=1 for n1 0 pa, Xfli (1)=P for n1= pa (p is prime). And so
R (Xn, xm 1)=0 for n1= d n =1,
R (Xn, x'n 1)=P
for n1=7 1 =r' R (X, xin 1) = 1 in all other cases.
(n0
(n)
828. It is obvious that R(Xn, X,) is a positive integer which is a divisor of R xm 1) and of R (X,, xn 1). Denote by d the greatest common divisor of m and n. If m is not divisible by n, and n is not divisible by m, then
and are different from 1 and are relatively prime. By virtue of the red d sult of the preceding problem, R (X,,, x'n 1) and R ( X xn 1) are in this case relatively prime, and therefore R X ,,,)= I . It remains to consider the case when one of the numbers m, n is divisible by the other. For definiteness, say n divides m.
If m=n, then R (Xm, X,,) =0. If 1 n -- is not a power of a prime, then

R(X,,,,xn-1)= land, hence, R R X)-= X,,) = 1. Finally, suppose that m= npA. Then v \ R JCS 1) k81.
8/n
275
m All factors on the right are equal to unity, except those for which is a power of the number p. If n is not divisible by p, then there is only one factor different from unity when 8=n and
4+ (m)
R(Xm, X,,)= R (Xm, x"

II 8=n and 8= -. Then
p
(n)
I) =p "(min) =p92(")
If n is divisible by p, then there are two factors different from unity: when
R (X m, Xn) =R
R (Xm,
o (In)
x nip _
x n 1 )
w (mln) w Imp/n)
P
cp (m) [
I I
(I) (m)
=P Thus,
, (p 1,
1,
=p
= pc - !n)
R (X ,, Xn)=0 when m=n, R (X m, Xn) =ps (n ) when m=np, R (X m, Xn)=1 otherwise. 829. (a) 49, (b) -107, (c) -843, (d) 725, (e) 2777. 830. (a) 3125 (b2 -4a5)2, (b) A4 (4A-27)3, (c) (b2 -3ab+9a2)2, (d) 4 (A2 8A+ 32)3. 831. (a) 2 ; (b) A1= 3, A2,3= 3 (
Ay= -3, A3=125;
2+ - 2
(c) )1=0,
7 2i V 3 . (d) Ai= -1, A2= -3A7,1- 2 9 832. In the general case, if the discriminant is positive, then the number of pairs of conjugate complex roots is even, if the discriminant is negative, then it is odd. In particular, for a third-degree polynomial, if D >0, then all roots are real; if D < 0, then two roots are complex conjugates. For a fourth-degree polynomial for D >0, either all roots are real or all roots are complex. For D <0 there are two real roots and one pair of conjugate complex roots. 833. f=xn+ a, f'=nxn (n-1)
R(f', f)=1." c,"-', D(f)=(-l)

10'
n"
276
834. f=xn+px+ q, f' = nxn-+ p,

n-2
R(f', f)= n" 27-c n -1 1
rj
k= 0
n-1
n-1
( q+
ek
13
where e =cos
i sin
27r n -1 P
=nn qn-1+(_ 1)n-1 (n _ On- pn,
(n-I) (n-2)
R(f', f)= nni-1
nn- 112n -1
(n-I) D(f)=( - 1)
nn q '
1)
(n-
tations: mi=
835. Let the greatest common divisor of m and n be d. Introduce the non ni = 71, e is a primitive nth root of unity, 71 is a primitive
nith root of 1, aox'n n aixin + 2= f (x). Then f ' (x) (m+ n) aox'n +MaiXm-1. The roots of the derivative are = 2 fl1+.k= -1/ Furthermore,
n- 1
ma, (n+n)ao
ek= m
sk, k=0, 1, ..., n-1.
R(f' f)=(m+ nr + ern qn _1

k= 0
[ a2+ na, m+ n na, n
c km]
in
n,-1
=(M -Fn)m-En
GT -E-n
[11 (
k= 0
612+m+
= (n + n)'n +n an+n o
-
2 [ anll'
1Yn1
nn, m' ' o i (m+nynl+anl
an gaT- I [(m+ nyni"=
a2 ''1 + (-
el 111'1
anh+nod
and, hence,
(m+n) (m+n -1)
D ( f)= ( - 1)
a3-1 a2-I [(m +n )ni +n,
a21
+(
)mi +II,. -1 n n1mm,. 071, + n,y1.
836. The discriminants are equal. 837. x1 x 2 +X3 x4 -x4 x3 -x2 X4= (X3. X4)(X2 -x3),
X3+ X2 X4 X1 X 4 x2XS =(X1 X2)(X3 x4), x1 X4+ X2 x3 x1 x2 X3
x4= (x1 X3)(X4 x2).
Squaring and multiplying these equations, we get the required result.
277
838. Let f (x)= ao (xx1) (x x2)

D (x) (xa))= xir (c1 x 2)2
(xxd. Then
. (a xn)a
i<k
(xixk)2=D (f (x)) [f (a)]2.
839. Let us denote cp (x)= xn 1+xn 2+ ...+1. Then (x-1) cp(x)=xn I, whence it follows that
(n-1) (n-2)
D(cp) [cp (1)]2 =D (xn 1)=( 1) Consequently,

(n I) (n-2)
D (cp)=( 1) +(a-1)xn a. Hence,

n (n-1)
nn- 2.
840. Let co (x)=xn+ axn l+can-2 + ... +a. Then co (x) (x 1)= xn +
an R On+ 1.2 n
ar+1.
(na + 1)2 D (cp)=( 1)
Thus,
D(p)=(
rz (n-1)
an-1
(n + on+). a + nn 0 _on+3. (1 + na)2
841. Let f (x)= ao(xx1) (xx 2) cP (4= bo Then

D (fip)= (ao born +2n-2
(x .)12)
(xxn),
(X Ym).
11 (xixk)2 11 (yiyk)2
<k i <k
n m X
i =1 k=1
(xiykr=asn
i<k
(xi_xk)2bir -2
fl
i <k
n m Y kr [a'ofl
b0
fl fl (xiyo] =D(f) D (9) [R (f,

1=1 k=1
p)]2
i (XP 842. X
mi
1)= XP 1.
M-1
Consequently,
777 -1
D (X ,n) D (xP
1) [R(xP
1, X pm)]z =
D (XP 1).
Substituting the values of the known quantities, we get

D (X )=pmP (m+0 13
rt
mi
(- 1)
1 Pm - 1 (P-1) 2
11 C) 1) 8
843. Xn rj (.0-1) k8= (X" 1)

6/n 6/n n
278
Let e be a root of Xn. Then

n '6; (e)= izen -1
n( 0_1) k 8 1.
8/n 8# n
To simplify computations, let us first find the absolute value of the discriminant of Xn :
I D (X01= n
(c) ,=0)(")
8/n 8# n a.
`1 0 _0) k8 1
( n) n
(n)11 [X n (1)]
Sin 80 n n 8
8
k
Now, X n(1) differs from I, provided only that w is a power of a prime

8
number. On the other hand, v. ( n ) is different from 0, provided only that
S is not divisible by the square of a prime. Thus, in the latter product we

must retain only those factors corresponding to =P15
P2,
pk , where
pk are distinct prime divisors of the number n. Pi, p2, Thus,
D (Xn)l Pin
ncp (n)
ps P (n)IP
Since all roots of X are complex, the sign of the discriminant is equal
cp (n)
to (-1) 2
. Finally, co (n) cp (n)
D (Xn)=(-1) 2
11 pcp (n)/p p/n

844. E=n! (1+
X
+ . ..+
n! ) x"-
x"
E,,=n! 1+1 + 1 (n-1)! )
279
Hence
En= En - x", R (En, En)= ( -x1)n- ( _ 1)n [(- 1)n nUn =(n!)n,
J=1
n (n- I) D (E,,)=(-1) 845. It is easy to establish that (nx+n -a) Fn - x (x+ 1) F;+ a (a -1) . . . (a - n) =O. n! a (a -1) . . . (a - n) n!
Cn
(nOn.
Let x1, x2, . , xn be roots of Fn. Then

Fn (xi) = xi (xi
where c-
Hence
R (F, F,;) flxi n (xi +
- a (a -1) . . . (a - n + 1)
n!
(a -1) . . . (a - n) n!
(a a+ 1)n-2 ( a _ n)n-i an-1(a-1)" 2 (a 2)n- 2 (on-2 n (n- I) D (En)= (-1)
an-1(a- i)n-2 (a -2)n-2 ...(a - n + 1)n-2(a- n)n(non-2 R (P,,, Ph)-=n"R Pn-1, -
846. Ph= nP,,_ /. Hence
Furthermore
xPn -i+ (n -1) Pn _ 2= 0. PnConsequently, Pn ()= -(n-1) P n _ 2 (0 if is a root of Pn_ 1, and therefore R (Pn, Pn-.1)= (- 1)n- 1 (1- 1)n-1R (P t,- 23 Pry-1) R (Rn_15 Pn_ 2). (n -
It is now easy to establish that

n (n- I) R (Pm n-1)= (-1) 2
2 1. (n - 1)n-(n -2)n-2 . . . 2
Finally
D (Pn)=1 22 3' 847. D (P,,)=1 23. 35 ... n2,2-1. 848. D (Pd=2" - 40. 849. D (Pd= (a +
2n(n-1).
(n - 1)n -
850. D (Pd =1 23 3'... en-1. 12 (n-1) 32(n-2) 851. D (P,,)=22 34 . .. on- 2 (n+
(2 n - 3)2.
852. Let f (x)= xn+ aixn
+ a,,=(x - x2) (x- x2) ... (x- xn).
280
D(f)=II (xi-xk)2. We seek the maximum of D(f) by the rule for finding a relative maximum by solving the system of equations x1+4+ +$2=n (n-1) R2,
a (
axi It is easy to see that
D (4+ .4+ . . . + 4)=0. aD _ Df" (xi) axi f' (xi)
We thus have
f" (x) D -2A x f (xi)= 0 for i=1, 2, ..., n. Thus, a polynomial f (x) that maximizes the discriminant must satisfy the differential equation
cf (x)- 2Axf '(x)+ Df " (x)=0

where c is some constant. Dividing by and comparing the coefficients of
x", we find that the differential equation must have the form
of (x)-xf ' (x)+ cf" (x)=0 where c' is a new constant. Comparing the coefficients of xn-- 1 and xn-2, we find ch= 0, a3 = n (n-1)
2 e. Now we can determine e. Indeed,
n (n-1) R2 =
x2+
+ =
2a2= n (n-1) c'
whence c'=R2. Continuing to compare the coefficients, we find that f (x) is of the form n (n-1)(n -2) (n- 3) R4x-_ n 4 f (x)=xn - n (n -1) R 2 xn- 2 + 2 24 It is easy to see that
f (x)= Rn P,, ( R x )
where P. is a Hermite polynomial.
D (n=Rn (n-3) . 1 . 22 . 33
n".
This is the desired maximum of the discriminant.

853. 22n (_ 854. mmn (_ 1) ao an [D (f))2.
n
in (in 1)
co n 1 an -I [D (f)]m.
855. F (x)= fl (cp (x) -xi).

1=1
Hence
D (F)-=
i =1
D (cp (x) -xi) {
fi
R (cp (x)- xi, cp (x)-xk
i<k
281
It is furthermore obvious that

R (p (x) xi, p (x) x k)= (xi x k)m .
Therefore
D (F)= =1 D (p (x) xi), 1 1 (xixk )2m = [D (f)]m 1 1 D (cc ,(x) i <k i=1
which completes the proof. 856. (y+1) (y-5) (y-19)=0. 857. (a) Solution. x3=3x +4. Let y=1 +x +x2 where x is a root of the given equation. Then
yx =x+x2+x3=x+x2+3x+4=4+4x+x2, yx2=4x + 4x2 + x3 = 4x + 4x2 + 3x +4 =4 + 7x+4x2.

Eliminating x, we get 1y 4 4 1 4 y 7 1 1 4 y
= 0,
Y3-9y2 +9y-9=0;
(b) y3-7y3+3y-1=0; (c) 3)4+ 5y3 + 9y2 + 7y 6 =0 ;

(d) y4 I2y3 +43y2-49y+20=0.
858. (a) y3-2 y2 +6y-4 =0, x
y2-2y +4
2
y2-3y+ 2
(b) y4- 9Y3-1- 31Y3- 45y+13=0, x
(c) y4+2y3 y2 2y+1 = 0, there is no inverse transformation.

859. y3 y2-2y+1=0.
The transformed equation coincides with the original one. This means that among the roots of the original equation there are roots x1and x2 connected by the relation x2 =2-4. 860. Let x2 = cp (x1), where p (x1) is a rational function with rational coefficients. Without loss of generality, we can take it that x2 =axl+bx1+ c. The numbers ax+bxi+c, a4+bx 2+c, axl+bx3+c are roots of a cubic equation with rational coefficients, one of the roots of which coincides with the root x2=a24+bx1+c of the given equation. Since the given equation is irreducible, the other roots must coincide as well. Thus, either ax3+bx2+c= =x3, a4+bx3+c=x1, or axl+bx2 +c=x1, axl+bx 3+c=x3. The latter equation is impossible since x3cannot be a root of a quadratic equation with
282
rational coefficients. Thus, on our assumption the roots of the given equation are connected by the relations
x2= ax?+ bx,+ c, x 3= ax2+ bx2+ c, x1=a4+ bx3+ c.

Consequently 1/.5 (x2 xi) (x3x2) (x1 x3)
= [axl + (b 1) xi+ [a.4 + (b 1) x2+ c] [ax2+(b 1) x3+ c] is a rational number, being a symmetric function of x1, x2, x3with rational
coefficients. The necessity of the condition is proved. Now suppose that the discriminant D is the square of the rational number d. Then
X2 X3 (X1 X2) (Xl X3)
3xT+ 2axi +b
On the other hand,
x 2+x3= ax1.
From this it follows that x2 and x3 are rational functions of x1. This proves the sufficiency of the condition. 861. (a)
2+1/2- + 1/ 6
4
3
3
(b) 3+71/
1/ 4
23
4 4 (c) 1+31/2 +21/2 1/8

(x
862. (a)
_f_ 1
, (b) 17a2 3x+ 55, (c) 3 10a +8a2 -3a3,
(d) the denominator vanishes for one of the roots of the equation. 863. mx2 i +nx1+p864. If
X2
(pm bm 2 amn n2) xl + (amp np crn2) mx1+ ma n

aX1
y xi + ' ax2 +p
yx3 +
then
x3
x1=
ax3 +P 1x3 + 8
(aa +PY)x2+(a+ 8) + 8) ix2+ (PY+ 82)
On the other hand, x1= relation
--8x2-FP , whence follows the necessity of the yx2 C a8 PY = (a+ 8)2.
283
865. Let clexn + (he' 1+ . . . + a.= a0(x - xi) (x -x2) ... (x-xn). Then aoxn- aixn-1+ +(- Dna.= ao (x+xi) (x+x2) Multiplying these equations, we get
(x+xn).
a0 (x2 -
(x2-4)
... (x2 -
x2 , i)
=(ao xn+ a2 xn-a+ . )2 -(a, x"- '+a3 x"-+ ...)2 From this we conclude that in order to perform the transformation y = x2, it is necessary to substitute y for x2 in the equation
(ao xn+a2 xn-2+ ...)2-(ai xn-i+a3 xn-3+ ...)2=0.

866. The desired equation results from a substitution of y for x3 in the equation
(ao xn+a2 xn -3+ ...)3+(ai xn-'+a4 xn-4+ ...)3 +(as xn-2+a, xn'+ ...)3 -3 (ao x"- a2xn-3+ ...) x(a,xn-'+a4 xn-4+ ...)(a2 xn-24-a5 xn-5+ ...)=0.
867. There are only a finite number of polynomials e+aix" + with integral coefficients the moduli of whose roots do not exceed 1, because the coefficients of such polynomials are obviously restricted :
ak I
n (n-1) k!
(n- k + 1)
Let f= xn + aixn- 1+ +an, an 0 0 be one of such polynomials and let x1, x2, ..., X. be its roots. Denote fm=(x- xr) (x- x2) (x-x All polynomials fmhave integral coefficients and all their roots do not exceed 1 in absolute value. Hence, there are only a finite number of distinct roots among them. Choose an infinite sequence of integers mo<m1<m2< ... such that fnm= f,-= f,= .... This signifies that
mi Xi = Xcc, mimo X2 = Xcc, ,
xn = x'n an where (oil, oc2, ..., an) is some permutation of the indices 1, 2, ..., n. Since there are infinitely many exponents miand only a finite number of permutations, there will be two (and infinitely many) exponents mc and nk to which corresponds one and the same permutation (cci, 0C2, ..., For such exponents we have the equations
284
which show that x1, x2, ..., xnare roots of unity of degree mil mil because x1, x2, ..., xnare nonzero, by virtue of the condition an 00. 868. Let F(x1, x2, ..., xn) be a polynomial which changes sign under odd permutations of the variables. Since F x2, x3, , x)= F(xi, x2, x2)=-0, F(x1, x2, ... xn) is divisible by x 1 x2. In similar fashion we prove that F (xi, x2, ..., x,;) is divisible by all the differences xi xk. Hence, F (xi,
x2, ..., xn) is divisible by A= 11 (xi xk) equal to the Vandermonde deteri >k
minant. Since the determinant A changes sign under odd permutations of the variables, is a symmetric polynomial. A 869. Let p (x1, x2, ..., xn) be a polynomial that does not change under even permutations of the variables. Denote by c-P (xi, x2, .., xn) a polynomial obtained from so (x1, x2, ..., x) by means of some definite odd permutation. It is easy to verify that for every odd permutation, (ID goes into :1), ci) goes into cp. Hence, qDP does not change under all permutations, cpcp changes sign under odd permutations. Next, cP=
<15 4-,P 2 (15 2 cP
= F2 - FF,
where A is the Vandermonde determinant. On the basis of the result of Problem 868, F2is a symmetric polynomial; F1is also a symmetric polynomial since it does not change under all permutations of the variables. 870. (f9f2) A, where f1f f2 are elementary symmetric polynomials in x1, x2, ..., xn. 871. u3+a (m+(.i+y) u2 -1-[(a2 +132 +y2)b+(ocfl+ay+3y) (a2 6)] u
c (G3+ (33+-M
ab 3c (op + 42+ ,2y ,y2 p 2y py2)

2
(Y a) V A 2 whore A is the discriminant of the given equation. 3pp' u 27 qq ' +V AA ' 872. u3 =0, where A and A' arc discriminants of 2 the given equations. 873. Let y=ax2+bx+c be the Tschirnhausen transformation connecting the given equations. Then, for some choice of numbering, . +4+4)+b (.4+.4+.4)+c (x1+x2+-x3) y1+x2Y2+x3Y2=a (4 will be a rational number. Hence, one of the equations
(a3 3a b + 6c)+
(C4 ( 3) (13
u3 3pp' u
27 qq'
2
AA'
(Problem 872) has a rational root. Whence it follows that 1/ AA' will be a rational number. This proves the necessity of the condition. Conversely, let the equation u3 3pp' u have a rational root u.
27 qq V
2
(*)
285
It is easy to see that the discriminant of equation (*) is equal to 272 (ql -1/ q' )2and, consequently, differs from -1/S, by a factor equal to the square of a rational number. Hence, the difference u' u" of the second and third roots of the equation differ by a rational factor from 1/A. For y y2, y3we have the system of equations: Y1+Y2+Y3=O, x1 Y1 +X2 Y2 +X3 Y3 =-14
(X2 X3) Yl+ (X3 X1) Y2 + (X1 X2) Y3=Li' 1/"=r1/A
From this system we find
-3ux1+ (x2 - x3) rirA

.V1=
6p
6, is expressed rationally in terms of x1. This proves the sufBut (x2 -x3) 1/: ficiency of the conditions. 874. The variables x1, x2, ..., xnare expressible linearly in terms of 7 11, 712, ,17,-1. Hence, every polynomial in x1, x2, ..., x may be represented as a polynomial in f .22
F (x1, x2, .
, x)=
7)T1 712 . . .
In a circular permutation of the variables xl, x 2, , xn, the monomial 71nn i I acquires the factor e- (0c1+ 22+ + (n 1)c(n 9. Hence, Aft 71Ti 712z so that F(x1, x2, ..., xn) should not change in circular permutations of the variables, it is necessary and sufficient that oc1+2a2 + + (n-1) an _ be divisible by 71. 875. We can takefl, 717, 712 7IF 2 , (n -1)
-
2 2 )2 /-I I/ 3 P2 where piand (1)2 are some rational functions of xl, Then )12 x2, x3with rational coefficients that do not change under a circular permutation of x1, x2, x3. It is easy to see that every rational function of x1, x2, x3 that remains fixed under circular permutations of the variables is expressible rationally in terms of f1=x1+x2 +x3, qh and 1P2. It is sufficient to prove this for )12 7)1 and M. But 1
7
876. Let 7h=xi +x2e+x3e2, 12=xi -Fx2 e2 +x3 e where e=
1 lrS +i
712 711 2= (P1+ jcP2 V 3

71! . )2 )12 712 = (
1)2
711.
=(cP1 -1- 42
V 3 )2(cpiicp21/ 3 ).
1X4
877. For n=4,

11=X1+ iX2
712 =X2. X2 + X3 X4,

13 = 7
iX2 X3+ iX4
II. 1215
86
PART III. ANSWERS AND SOLUTIONS 711 712

7la
01, 0 2, 03 are rational 171 functions (with rational coefficients) of x1, x2, x3, x4that do not change under circular permutations. It is easy to see that together with f=x1 -Fx2 +x3+x4 they form a system of elementary functions. Indeed, Ut 01=7)113, 02+iO3 = , 01iO3=
713 7i2
'127)
2 -
02 - iO3 e2 jo, 3"2 -3 01 ' 7 ' '1 _ - 01 (02 + i0,

7)1=
0; . (02 (00 02 - iO3
878. Let
+x2e2 +x30+ X4E 4 + X5, 12=x1e2 +x2c4 +x3c+x4e3 +x5, 733=x1s3+x2e +x3e4 +x4e2 +x5,
7)4 = X1C4 + X2e3 + X3e2 +x4E+ X5.
731. Let us consider the rational function Ai = and arrange it in po-
wers of e, replacing 1 by e 0 0 e4 : Al.= EP' + C2(1)3+ e3T3+ E4(P4 The coefficients of m co 22 ao co 3, 4 are rational numbers. Substituting 0, 0 and 0 for e, we get
7)04 = pi + E4 731
cq),-F Cpq, cp4,
A3
7)311 =C3 pi4_ cp2 + p3 +

7)4
A4 = 7)413 =
2(p3 + Ecp4 (pi + E3 + E
A2, A3, A4
For the "elementary functions" we can take f, P2, (P3, cp4. Indeed Al, can be expressed rationally in terms of them. Furthermore, . 12 .71r 2 = A ri A2Az1, 14 y]i4=Ai2 AlA3 I x47 1 ,
74-3 = A F 2 AzFl,
71,=AIA2 IA3 Ai

879. (a) Dimensionality r=2, the basis is generated, for example, by X1and X2; (b) r= 2, the basis is generated, say, by X1, and X2; (c)r= 2, the basis is generated, say, by X1and X2. 880. (a) The dimensionality of the intersection is equal to one, the basis vector Z= (5, 2, 3, 4)= X1-4 X2 = 3 )72.
287
The dimensionality of the sum is equal to 3, the basis is generated, say, by the vectors Z, X1, Y1. (b) The sum coincides with the first space, the intersection with the second. (c) The sum is the entire four-dimensional space, the intersection consists only of the zero vector. 1 1\ 5 1 1, 0), (b) (1, 0, 881. (a) (4 4' 4 4 )'
882. (a) xl= x,
+X2
2
X3 -
,
X4 =
x1
x2 + x3 2
.7C4
5
x1 -x2 -x3+x4 2 '
, xi+x2-1-x3+x4 2 x' 3 =x4, x4 =x1x2+x3 --x4
(b) x' 1 =x2 x3 +x4, x' 2= , 1 883. x',..4+.x;x4= 8 884. Let ao + aicos x+ a, cos' x+ Then ao=bo b2+b4 . ,
a=
+ ancosh x= b o + bi cos x
+b 2cos 2x+ + b n cos nx.
2k {bk
(k+2p)(k +p-1)(k+p 2) .. (k+1) ,

nk 1, c.p. 72
ok+2],
b0 = au+
E
15p5 2
2 -2P
C1pci2,0,
bk = 2,_k (ak+ I
nk
2-2p
cP k + 2pak+24
885. The point of intersection with the first line has the coordinates 7 11 ) /14 1 , with the second, the coordinates (42, 1, 7, 11). \3' 9 ' 9 , 9 886. The straight lines Xo+ tX1, Yo + t 1' 1lie in the manifold Xo+ t (Y0 X0)+tiXi + t2 Yl 887. For the problem to be solvable for the straight lines X0+ tXi, Yo+ +tY 1, it is necessary and sufficient that the vectors X0, Yo, X1, Y 1be linearly dependent. This is equivalent to being able to embed straight lines in a three-dimensional subspace containing the coordinate origin. 1 +t2 Y2 can be embedded 888. The planes Xo+t1X1+t2X2 and Y o +t1l' in the manifold X0 + t (Y0 X0)+4 Xi + t 2 X2+ ts Yi+ t44 889. There are 6 such cases; II*
288
(1) the planes have no common points and cannot be embedded in a four-dimensional linear manifold (the planes cross absolutely); (2) the planes have no common points, are contained in a four-dimensional manifold, but are not embedded in a three-dimensional manifold (they cross parallel to a straight line); (3) the planes have no common points and are embedded in a three-dimensional manifold (the planes are parallel); (4) the planes have one common point. In this case, they are embedded in a four-dimensional manifold, but cannot be embedded in a three-dimensional manifold; (5) the planes intersect along a straight line; (6) the planes are coincident. In three-dimensional space, only cases 3, 5, 6 are realized. 890. Let Q= X0+ P be a linear manifold, let P be a linear space. If A', E Q and X2 EQ, then Xi= Xo + Y1, X2 = X0+ Y2 , where Y, and Y2 belong to P. Then aX1+ (1 a) X2 = X0 + Yi + (1 a) Y2 E Q for any a. Conversely, let Q be a set of vectors containing, together with the vectors A'1, A'2, their linear combination X1+(1 a) X2 for arbitrary a. Let X, be some fixed vector from Q and let P denote the set of all vectors Y= X X0. If YEP, then c YeP for any c, because c Y=- cX+ (1 + c) X, X,. Furthermore, if X,EP and Y2 = X2 - X,EP, then a Y1+ (1 a) Y2 = aXi + (1 a) X2 - X0 eP for any a. Now let us take some fixed a, a00, a0 I, arbitrary c1, c2. Then Cl c2 Y2eP for any Y1, Y2 EP, and, hence, also a / 1a c2 Y2EP. Yi + (I a) 1 ci + c2 Y2= a ca
,
Consequently, P is a linear space and Q is a linear manifold. Remark. The result is not true if the base field is a field of residues modulo 2. 891. (a) 9, (b) 0. 892. (a) 90', 893. cos (p = (b) 45, (c) cos cp 1 77
= n 8 5 , cos C= 2 894. cos A= , cos B= V39 3 V78 895. V n. 896. For odd n there are no orthogonal diagonals. For n= 2m, the num27,_1 1 . ber of diagonals orthogonal to the given one is equal to C' 897. The coordinates of the points are represented by the rows of the matrix 0 0 1 0 0
/ 3 4
0 0
1
0 0
1 2
1
1 /' 4 V6 = 1/ I 2 1 V12 1 V24
r:+ 1
2n
V20 (0 1)
7. LINEAR ALGEBRA CH .'
289
898. R=V 2 (nn i) The coordinates of the centre are 1

(
2'
1 1/12' 1 15'
' 2 15'
1 1/2n (n -1)' 1 1/15/
1 1/ 2 0+ 1) n
899
3 1/ 15'
900. (0,
1 1 v2 v2 , - , 0)
901. For the other two vectors we can take, say, 1 1 (-13, 5, 6, 2). (0, -4, 3, 1) and 31/26 1/26 902. (I, 2, 1, 3), (10, -1, 1, -3), (19, -87, -61, 72). 903. For example, 7 3 -4 -2 -29 5/ (3 9 - 37 51
904. The system is interpreted as a problem in seeking vectors orthogonal to a system of vectors depicting the coefficients of the equations. The set of sought-for vectors is a space orthogonally complementary to the space generated by the given vectors. The fundamental system of solutions is the basis of the space of the desired vectors. 905. For example, 1/1 (1, 0, 2, 906. (a) X' = (3, 1, -1, -2) EP, (b) X"=(2, 1, -1,4) IP, 1), (1, 12, 8, 17). 1/498 (1, 7, 3, 3) eP, X"=( -4, -2, 6, 0) IP.
A. be linearly independent, and let P be the space 907. Let A1, A2, spanned by them. Furthermore, let X= Y+ Z, YeP, ZIP. Set Y=ci Ai+C2 A2+ ... +cm Am.
Form a system of equations to determine c1, c2 , ..., cm ; to do this, use scalar multiplication of the latter equation by Ai, i=1, 2, ..., m and take into account that YAi= XAi. We obtain c2A1A2 -1- cmAiAm = XA D ci A,Ai + cz A-1- + crnA2Am= X A2.
CiAmAl+ C2A M A2+
+ CniA
i = XAm.
290
The determinant A of this system is nonzero by virtue of the linear independence of Al, A2, ..., Am. cmand substitute them into the expression for Y. This We find c1i c 2, yields 0 1 Y= -4
X A, XA 2 X Am
.
A1
.Al A2A1
A2
A 1A2 /4
Am
AiAm A 2,4 m
Ain
2 A mA 1 A,A
and
X A1 X Z= A2 X Am X
Al A2

Am AiA m A2A m
in
Al A2 A1 AmA1
A1A2 A2 A mA2
These equations are to be understood in the sense that the vectors Y and Z are linear combinations of the vectors in the first row with coefficients equal to the corresponding cofactors. From this we finally get 0
Ai A2A1 AmA1 A 1A 2 .q A mA2 X Am A iA nz A2/I m Am 2
A1 X
Y2 = Y(XZ)=- YX= A
A2 X Am X
and
X2 41 X XA 1 Al A2A1
AmA2
XA2 A 1A 2 Al /
m A2
A A2Am
Z2=Z (X Y)=ZX=
A2 X Am X
A%
908. Let Y be some vector of the space P and let X' be the orthogonal projection of the vector X on P. Then cos (X, Y)= XY IX11 Y1 X' Y 1X11 171 1 X'l I Y 1 cos (X', Y) 1 XI .1 Y1
X'i
I X1
cos (X' Y)
whence it follows that cos (X, Y) attains a maximum value for those Y for which cos (X', Y)= I, that is, for Y=aX' when a >0. 909. (a) 45, (b) 90.
291
910. t 911. 1 X Y12 =1 (X X')+(X' n12=1 X X' 12+ IX' Y 12?. 1 X X' 1 2. The equation is possible only for Y= X'. 912. (a) 1/ 7, (b) 913.
/7 !
(n + 1) (n + 2) ... 2n-/ 2n + 1
914. The required shortest distance is equal to the shortest distance from point Xo Yo to the space P+ Q. 915. Let one of the vertices lie at the origin of coordinates and let Xi, X2, ..., Xn be vectors emanating from the origin to the other vertices. It is easy to see that A1=1, XiX; = . The manifold passing through the first m+1 vertices is a space t1 X 1+ + t. X.. The manifold passing through Xn). the other n m vertices is X,,+ t,1 (X,+1X) + + t,,_ The desired shortest distance is the distance from Xn to the space P generated by the vectors X1, X2, Xm, Xn Xrn -1- 11 Xn Xn _ 1. Let Xi + ... +t. X,,,+t, 4_1(X Xm+ 0+
+tn _ i(X,,
where YIP. Forming the scalar product Xn by X X n X,n +i, Xn Xn _ 1, we get the following system of equations for determining t1, tn _i:
ti + 2 t 2 + + 2 tin= 2 ,fin+1+ -2 n2+2+

1 ti +t2 + 2
1 ti+ 1
1 1 2 tnl 2 1 1 1 tm i+tm + 2+ +1 = 2- to -1 2 1 1

1 1 1 t2 + +tm=' 2 tm+1+ 2 tm+2+

tm 1 tin+i tin+ 2 =
+ t 1 2 '
1
whence t1=t2= Consequently Y-
M+1
=- to 1= - -
n m
Xn7+1+ Xin+ 2 +
X, ' +
+ X2+ ' + Xm
nm
in+ 1
Thus, the common perpendicular is a vector connecting the centres of the chosen faces. The shortest distance is equal to the length of this vector
)(m+1) m +1 YI= V2 (n-1'
292
916. (a) The projection of the vector (t1-F2t2, t1-2t2, 4+54, t1+20 on the first plane is (t1+24, t1-2t2, 0, 0). Hence, 2a+8t1 2A2 +8 cos2 (p 4a+ 14t1t2 + 37q 4A2 +14X+ 37' 8 tz where A= . This expression attains a maximum equal to for ) = 4. (b) The angle between any vector of the second plane and its orthogonal 4 . projection on the first plane remains fixed and is equal to 1r 917. The cube is a set of points whose coordinates satisfy the inequalia y, 1=1, 2, 3, 4 . Here, a is the length of a side of the cube. ties We pass to new axes, taking for the coordinate vectors, e=( 1 1 1 2, , 1 1 1 2, (1
2'
1 1 1 1 ,2 -, ,
/'
and e,',= I1 1\
- 2' 2 ).
These vectors are orthonormal and their directions coincide with the directions of certain diagonals of the cube. The coordinates of points of the cube in these axes satisfy the inequalities 4+ 4+ 4x'4 -a, 4+4 x4 ...a, 4 - + x4 a. We get the intersection that interests us by setting x' i =0. It is a solid located in the space spanned by e2. e3, e4 '. and the coordinates of the points of which satisfy the inequalities + x2 + x; + a. This is a regular octahedron bounded by planes intercepting segments of length a on the axes. B! 918. V' [Bi, B2, . . , Bn]= B2B3 131132 . . . BiBm M B2Bm
1 B,1B1 BmB2 . This formula is readily established by induction if we take into account the result of Problem 507. Frcm this formula it follows immediately that the volume does not depend on the numbering of the vertices and that V [CBI, B2, Bm] =ICI V [Bi, B2, Now let Bi=B;. +BL C1, CI, C7 [be [the orthogonal projections of the vectors Bi, B; and B; on the space that is orthogonally complementary to (B2, ..., Bin). It is obvious that C1=Ci+C7. By definition, B]= V [B1, Bz, BA= I Ci I V [Bz, Bm], V [13;, B2, Bm1=-1 C;' . 1 V [B2, Bm]. =I C;1 V [B2, ..., Bm], V [X, .132, Since I Cz. I I Ci I+ I Ci I, it follows that V [Bz., B2, [B;, B2, [B", B2, . and CC equality sign is only possible if C; are collinear and in the same direction, which, in turn, occurs if and only if B' lie in a space spanned by B1, Bz, /3,, and the coefficients of B1 in the expressions of Bi , B'1 in terms of B1, B2, ..., .13mhave the same signs;
293
lie "on one side" of the space (B2, .. B.) in the space (Be that is, B.). Bz, Bi B1B2 . . . 2 B1 Ba . . . B2B,, BB I ---- I B Io. 919. V2 [B1, B2, Bn] = B Bn B, B,,B2 . . . Bn where B is a matrix whose columns are the coordinates of the vectors B1, 132. 920. The following two properties of volume are an immediate consequence of the definition: -= V [B B2, . Bm] (d) V [B2 + .X2 By, ..., B.1.B2,
for any X belonging to the space (B2, ..., Bin) because the points B1, Bi + X are at the same distances from (B2, Bm). (e) V [B1,
B2, ...,
I Bl I V [B2, ..., B
This follows from the fact that the "height", that is, the length of the component of vector B1orthogonal to (B Bm) does not exceed the length of the vector B1itself. Now let C1, C2, C. be orthogonal projections of the vectors B2, B2, B,,, on the space P. Assume that the inequality V [C2, Cm],V [B2, Bm] has already been proved. Denote by g the component, orthogonal to (B2, Bm), of the vector B1, by C; its projection on P. Since 15.;- Bi c (B2, ..., B,), we conclude that C- C (C2, ..., Cm) and, hence, that V [C1, C2, . C ]= V [C;, C2, ..., Cm] C; . I V [C2, ..., C77]. But obviously, I B' iI and, by the induction hypothesis, V [C2, ..., [B2, . ., C1 Bm]. Consequently, V [C1, C2, Cm] 131. I V [B23 , Bmi = V [B2, B2, ..., Bm ]. There is a basis for induction, since the theorem is obvious for one-dimensional parallelepipeds. 921. From the formula for computing the square of a volume it follows that V [A2, ..., Am , B ..., Bic ] = V [A ..., A] V [B Bk] if each vector Ai is orthogonal to every vector B ./. In the general case, we replace the vectors B1, Bk by their projections C Ck on the space that is orthogonally complementary to (A1, ., Am). By virtue of the result of the preceding problem, V [C1, Ck]-.c.. V [B1, . Bic], whence Ck ] [212, ..., Am , B1, ..., Bk1= V [A1, ..., Am , C1, = V [A1, ..., Am ] V [C Ck ]--C, V [Ai , ..., A,,,] V [B Bic ]. The content of this problem coincides with that of Problem 518. 922. This follows immediately from the inequality V [Ai , A m] , C..I A1 I x x V [A Am], which, in turn, is an immediate consequence of the definition of a volume. This problem coincides, in content, with Problem 519. 923. A similarity transformation of a solid in n-dimensional space implies a change in volume proportional to the nth power of the expansion factor. For a parallelepiped, this follows immediately from the volume formula, for any other solid, the volume is the limit of the sum of volumes of the parallelepipeds. Hence, the volume V2(R) of an n-dimensional sphere of radius R is equal to Vn(1) Rn. To compute Vn (I), partition the sphere by a system of parallel (n -1)dimensional "planes" and take advantage of Cavalieri's principle. Let x be the distance of the cutting "plane"fromthe centre. The section is an (n - 1)-dimensional sphere of radius 1/1 -x1
294 Consequently,
n-1 Vn (1)=2
0
f Vn _1(V1-x') dx=2
1 n _i
(1) f (1-x2) 2 dx
0
=Vn_3(1) f t 2 (1 -t) 2dt

0
= V ,,_ (1) B (1111 2 -j, 2) 1 (n 4 ) r
(1
vn_i(i)
From this it follows that
n
( 1.+1 )
v(1)_
T
+1)
(2
924. The polynomials 1, x, xn form the basis. The square of the volume of the corresponding parallelepiped is 1 1 1 2 n+1
1
1
3
1
n+2
[I! 2!
2
1
(n + 1)! (n + 2)! . . . (2n + 1) !
1
2n+1
n+1 n+2
925. (a) Xi =1, Xi =c (1, -1); A2 = 3, X2=c (1 , 1); (b) Al = 7, X1 =c(1, 1); A2 = -2, X2=c (4, -5); (c) Al =ai, X1= c (1, i); Ay = -ai, X2= c (1, -i); (d) Al = 2, X 1=c1(1, 1, 0, 0)+c2 (1, 0, 1, 0)+c3(1, 0, 0, 1); A2 =-2, X2=c (1, -1, -1, -1); (e) X =2, X= ci (-2, 1,0) +c2 (1, 0, 1); (f) X =-1, X=c(1, 1, -1); (g)A1 = 1, X1 =c2(1, 0, 1)+ c2 (0, 1, 0); A2= -1, X2 =c (1, 0, -1); (h)A1 =0, Xi =c (3, -1, 2); A2,3 = V -14, X2,3=c (32V -14, 13, 2-T- 31/-14); (i) Al =1, Xi= c (3, -6, 20); A2 = -2, X2=C (0, Op 1); 1, 1); A2=e, X2=C (3+2e, 2+3e, (j) X1=1, X2=C 3+3e); Xs= E2, Xa =c (3+2e2, 2+3E2, 3+30) where
1 C=
295
926. The eigenvalues A -1are reciprocals of the eigenvalues A. Indeed, from I A-1 A E 1=0 it follows that I E AA I = 0,
A . E =0.
927. The eigenvalues of the matrix A2 are equal to the squares of the eigenvalues for A. Indeed, let Then A AE I= (Ai A) (A2
(An A).
I A -0 (A2+A) (An+ A). Multiplying these equations and substituting A for A2, we get A2 AE 1=(A1 A) (AZ A) (A2 ,i A). 928. The eigenvalues of Am are equal to the mth powers of the eigenvalues of A. To see this, the simplest thing is to replace in the equation I AAEI= (A1 A) 0,2
A by Ae, Ae2,
(An
AEn I, where e = cos
n multiply the equations and substitute A for An. 929. f (A)= bo(A ) ( A ciE), hence If (A)1 = .1 A U.1 ... I A 930. Let I=
. . 27c sin ,
F (U F (Em).
(An A)
F (A) = I A XE I= (Ai A) (Az A) and Then

f (4= bo (x (x 2) (x
n I f (A) I =1)!
i=I
k= I
k)= f (A1)J (A2) f (),n).
931. Put cp (x)=f (x) and apply the result of the preceding problem. We get f (A) A E I= (f (f (f (A) whence it follows that the eigenvalues of the matrix f (A) are f (AO, f (A2), . , f (An). 932. Let X be an eigenvector of the matrix A corresponding to the eigenvalue A.
Then
EX= X, A X =XX , A2 X=X2 X, AmX= AmX.
296
Multiplying these vector equations by arbitrary coefficients and combining, we get for any polynomial f
f ( A) X = f (A) X i.e., Xis the eigenvector of f (A) corresponding to the eigenvalue f (A) .
933. The eigenvalues of A2 are n and n of ,-multiplicities
n+
2
,
1 and
n-1 respectively. Hence, the eigenvalues of A are +1/ n, 1/ n + / 17i 2 +1 and 1/ni. Let us denote their multiplicities by a, b, c, d. Then a+ b=-n - -2 n 1 c+d= . The sum of the eigenvalues of a matrix is equal to the sum of 2 the elements of the principal diagonal. Hence [a b + (c d) i] 1/n=1+c+e4 + +c`n-1)2. The modulus of the right side of this equation is equal to 1/72 (Problem 126). Consequently
(a b)2+ (c d)2= 1 .
Since the numbers cd and c+d are of the same parity, we conclude that n2 1 ab =0, cd= + 1 if is odd and
a b= +1, c d=0 if
Hence, for n=l+ 4k
n- 1 is even. 2
c=d=k, a=k+l, b=k or a=k, b=k+1

for n=3 +4k
a=b=k+l, c=k+l, d=k or c=k, d= k+1.

Thus, the eigenvalues are determined to within sign. To determine the sign, take advantage of the fact that the product of the eigenvalues is equal to the determinant of the matrix. Using the result of Problem 2c,9, it is czsy to find that for n=1+4k
a=k+1, b=k,
for n=3+4k
c=k+1, d=k.
Thus, the eigenvalues are completely determined. = +,(n 934. 1 + e e4+ n for n= 4k +1, 1 +e-1-e4 +
-Fc(" ')'= +il/n for
n=4k+ 3.
935. (a) Put L c =a". Then

Xk=Y
aek at" 1 aek
where ek = cos
2kn
i sin , k=0, 1, ..., n-1. n
2kn
CH. 7. LINEAR ALGEBRA (b) 4=a1+ a2ck + a3q+ (c) Xk= 2i
297
. . +an< I .
cos
km. 2. . . . n /1+1' k 1, anB )E,, a,,B a22 B).Ent . . .

an2B
936. aiB azn B ) A x BA.Emn= ( 021B

an2B
. . an,,B AErn
whence it follows, by the result of Problem 537, that where I A x BAE,,,i= Ip (B) (x) = I Ax AE,, I= By the result of Problem 930, cp (B) I=
11
i-1
n
(.ix-A).
n
k=1
ni
cP (PO= 1
i=1 k=1
(2(ik
Thus, the eigenvalues of A x B are the numbers cok, where oci are the eigenvalues of A, and P kare the eigenvalues of B. 937. If A is a nonsingular matrix, then I BA AEHIA (ABAE) A 1=1 A -1 1 I AB AL' I I A 1=1 ABM. I. It is possible to get rid of the assumption that A is nonsingular by passing to the limit or by using the theorem on the identity of polynomials in many variables. It is also possible, using the theorem on the multiplication of rectangular matrices, to compute directly the coefficients of the polynomials I ABA E I and I BAAE and satisfy yourself that they are equal. 938. Complete the matrices A and B to the square matrices A' and B' of order n by adjoining to A n m rows and to B nm columns made up of zeros. Then BA=B'A' and A'B' is obtained from AB by bordering with zeros. Using the result of the preceding problem yields what we set out to prove. The solutions of the Problems 939, 940, 941 are not unique. The answers given below correspond to a transformation that least of all deviates from the "triangular". 939. (a) ,j2+ (b) 42 +42 + x' 32, -=
x2 = X2 2x2,
x',=xi+x2,
-=- X2 + 2X2,
X:'1=
x3,
x:',=x,
+x,
298 (C)x12
PART
III. ANSWERS AND SOLUTIONS
x22 x?, 1
(d) Xi = + x3 x4,
,
1 xi= x1+ - x2+ x3, 2 2

,x;= 1
x1 2 x2,
x 2=x2 + x, +x4
x,,
x3 =x2 x3 +2x4 , x,',=
(e) x' 12 -42 +x,2 x42, 1 2 x,=x2+- x 2 x2= 2 , 1 1 x2+ x2 2 2 x4' 1 3 -2 x4 x . 3 4 940. x',2 -1- 42 +- 42 + 4 6
x4= 2
,
n+1 x'2 2n "
The variables x' x2,...., x. with the matrix
, x;, are expressed linearly in terms of x
0 0 0 941.
0 0
1 0
+Xn
0 ... 1 ( xi za
( x,x2 +x3 +x4 + 2
1 x4 + + x)2 2 (x,+ 2 3 (x + x + . + 4 4 35 3 xn
n 1
2
2) xn.
2 (n
942. The matrix of a positive quadratic form is equal to AA, where A is a nonsingular real matrix carrying the sum of squares to the given form, The positivity of the minors follows from the result of Problem 510,
299
943. Let f=a n Al+ . +ainxixn

+anixn xi+ . . . +annxn be a quadratic form. We denote f(k)=a 1,4+ +ak ixkx j + +aikxixk +akkxt,
al l ail(
Ak=
, r is the rank of the form f
ak 1 akk
Let
f=oc1x12 +0c2X?+
where
xl= XI+ bi2X2
.X2 = + + binXn, X2+ b2nXn Xn.
X'n =
Since the determinant of a triangular transformation is equal to 1, Di= =An=ociaz . an. Putting
Xk
= = Xn = 0
we get
f (k ) = ix( ik)2 cx2Xk)Z ak4 k)
where
Xlk) = +1)12)(2+ bikXk, X IC) = X2+ . + b2kXk.
(k) Xk =
Xk.
Whence it follows that Ak =ccia2 ock and that a necessary condition for the possibility of a triangular transformation to diagonal form is 41 00, 42 00, ..., 4, 00. It is easy to verify that this condition is sufficient. Furthermore,
ak= Ak A
for k r, ak=0 for k > r.
The discriminant of the form

fk (xk +1, , xn)=f CCIXi2
OCkX; c2
'2 =ak ix':+i+ . +anxti
is equal to ak i ak +2
a= Ak
An
944. The necessity of the Sylvester conditions was proved in Problem 942. The sufficiency follows from the result of Problem 943.
300
xn. Transform the 945. Let 1 be a linear form in the variables xl, x2, form f by means of a transformation with determinant unity, taking the form 1 for the last of the new variables. Then perform a triangular transformation of the form f to canonical form. The form f becomes ... an x; f= Gclxi2 +oczx'22 and xn =I. The discriminant of the form f is equal to a1a2 ... an. The discriminant of the form f+12is equal to oc loc2 . a n _ i(an + 1 ). It is greater than the discriminant of the form f, since all the coefficients a1, a2 , , an-1, anare positive.
946. f (xi, x2, . xn)
= ax1+ 2a22x2x2+ . . . + 2anixixn+

azl =a11 (x,+ x2+ ... .+ xn) 2+ (x2, , all an
xn)
where
f = cp
a21 and ( x2 + . . . + xn)2. all all
The form f1is positive and its discriminant is equal to
Diall
where Df is the
Df
discriminant of f. On the basis of the result of Problem 945, Df1> which completes the proof. 947. The proof is the same as for the law of inertia. 948. Form the linear forms /k = u1-1-u2xk + +u4-1, k=1, 2, . ,
are roots of the given equation. where xl, x 2, To equal roots will correspond equal forms, distinct roots are associated with linearly independent forms, real roots are associated with real forms, and conjugate complex roots are associated with conjugate complex forms. The real and imaginary parts of the complex form /k = ), +12 ki will be linearly independent among themselves and also relative to all forms corresk. ponding to roots distinct from xk and x' Form the quadratic form
n
f (a1,
, un)=.
k=1
u2 + u2 xi + . . . + u x7 1 )2
The rank of this form is equal to the number of distinct roots of the given equation. The matrix of its coefficients is S1 . Sn -1 (S1
Sn
S2 . . . Sn
_ 1 Sn .
5 211- 2
The sum of squares of the conjugate complex linear forms /k =Xk + + ip.k and 4= AI, ip-k is equal to 2A2 k -2112. Hence, the number of negative squares in any (by the law of inertia) canonical representation of the form f
301
is equal to the number of distinct pairs of conjugate complex roots of the given equation. 949. This follows from the results of Problems 948, 944. 950. The operation (f, cp) is obviously distributive. It is therefore sufficient to carry out the proof for the squares of the linear forms. Let
f= (ix,. + c4.2x 2+ ... + c(xd2, ? = (Pixi+ P 2X 2 + +13X71)2.

It is then easy to see that (PO= (cciPixi +(x 2P2x2+ ... + oc,,(3,,x,,)2 .- O. 2 1 - x2+ 3 - x3 , 951. (a) 4.x',2 + 242 2x 2, xl= x, -3 2 1 2 3, x2= 3x1+ 3 x2 x i+ 3 - x2+ --a x3; x3 = 3 x (b) 1 2 , 2 ax, a X2 - 3 -X3, 2.x? .42 + 5x32 , xi = -2 2 1 .4= d xi+ x2+
X3,
x;=
(c) 7x',2 + 442 + x;2,
1 1
2 2 xi x2 + -j x 2; 2 1 2
3 - X2 - 3 -X3 , X1. = -s- Xi + -
x2= s xi+ -a X2+ X3 o x; = x1+ s x2+ --33o

(d) 10x,2 + x22 + x;2, 2 2 1
x',= 3- x, + s x2
2V 5 V5
x 3,
x; = 5 x l 5 X2 1
2V5 41/ 5
75 3 '
"X3 =
(e)
7x',2 + 2x;2 + 2x 2 XI
= s x1+
15 xl+ 15 x 2 2 1
x2 s X3 ,
2 1/ V x2 = 5x, 55 x2, 21/5 41/5 xi + 15 15

X2 + -
V5 3
''
302
(f)
i+ 2.42 +54+84, x=.a x

X2 = s
1 2
2
x2+ x3,
2 s X2 x3,
Xy
x3
2 2 =3 x, 3
1 3x3:
(g)
74-24+74,
X2
22
, 2
x,1
22 x,,
2 X3, 3
3x2+
x,+-s x2+
2V2
X3 = - X
, 1/ 2
1/ 2 --6 - x3;
(h)
2 2 3x 1 3 1 1 xi2+ 5x? x;2, xi= 2 1 1
x2 x3,
a Xy 4 = -a + 2
2
X3,
x3= 3- xi + 3 X2 x3
(i)
2 X? ? X+ 342 +5X1,
X; =
1 XX2 y -2 X+ 3 x4 I
1 - Xs X2 = Xy -F 2X2 2
x4,
1 1 , 1 1 4 2 x i -2 X2+ -2 X3
X4= 9 y 2Xy
1 2 X3+ 2 X4
x12 +x22 -42 -4,
x1 2 x1+ V 22 x2'
x2=
V2 V2 2 x3+ 2 x4,
2 Xi - 2 x2,
-V
x3=
, 1/ 2
1/ 22
1/ 2
2 xi'
303 V2 4, 2 x
(k)
+ 3x?
x, 42 ,
V2 = 2 x2
V2 , 2 2, x2 = 2 x i+ 2 x ,
X3
1
2
1 1 i + 2 X2- 2 x
1
X3 '2 X4,
,
xi=
1 1 1 2 xi+ 2 x2+ 2 x2-1
41
(I)
x12 +.1- 2 +
4 2, 3x'
2 2, 2-2 x i + V2 x x'= ,
x2=
,1/
V 2 1, 2-- x3 + 22 x 1 1
1 X2 + 2 X3- 2 x2= 2 1 x 2 X4
=
.f. (m) X? 7 X 2 2 - 3X7,
1 I I x 1 2 x2 -2 x3+ x4;
1 1
4
1 , Xi =
7c1+
X2 +2 X4+
, -= X2
1 1 i_ 1 x3 21 xi ' 2 x2 2 -2 x4
1
x2= 2
1 1 1 xi 2 x2+ 2 x2 2x4 ' 1 1 1 2 x3+ 2 x4; xi 2 x2

'
1
xl= 2
1 x3+ 2 1 x4, 1 x2+ 2 1 xi + 2 42 , xi _ 2 2 2 + 3x,2_3x, (n) 5xi2 _ 5x, 1 1 , 1 1 2 x2 2 x2 2x4 ' x2 2 xi+ 1 1 1 , 1 2 x2 + 2x3 2 x4 ' x3 2 xi , 1
x4= 2 xi
, 1 952. (a) ill+ I A-'2 4. 2 ,X22 + X32 + 2
1 1 1 x2 2 x3+ 2 xl. 2
+x'2);
(b)
/ 2 11+1 ,2 1 + + x',f) 2 x1 2 `x2 +x32
304 where
PART Ill. ANSWERS AND SOLUTIONS
xl= l -- (xi + xo +
Vn x;=,xiixi +ai2x2+
+ xn);
+ ccinx, i= 2, ... , n
where (ail, ai ..., aid is any orthogonal and normalized fundamental system of solutions of the equation xi +x2 +
953. x;2 cos
-Fxn =0.
+ ... +x`"2 cos n+1 n +1 954. If all the eigenvalues of the matrix A lie in the interval, [a, b], then all the eigenvalues of the matrix AAE ate negative for A> b and positive for A <a. Hence, a quadratic form with matrix AA.E.is negative for X> b and positive for < a . Conversely, if the quadratic form (A XE)X X is negative for A> b and positive for A <a, then all the eigenvalues of the matrix A AL' are positive for X < a and negative for X> b. Consequently, all the eigenvalues of the matrix A lie in the interval [a, b]. 955. The following inequalities hold for any vector X: aX X..< AIX X, bX X<BX X<..dX X
n+ 1 + x? cos
rc
whence (a + b) X X <( A+B) X X-. .<(c+d) X X. Therefore, all the eigenvalues of the matrix A+B lie in the interval [a+ c, b+d]. 956. (a) This follows from the result of Problem 937. (b) AX12=AX AX=X A AX.< 1 X12 HA I12. The equal sign occurs for the eigenvector of the matrix AA belonging to the eigenvalue II A112. (c) 1 +B)X I<IAX1+1BX1-< (I1 A II + II B II) 1 X I for any vector X. But for some vector X0,
I (A + B) x I= (II A + BID I Xo I.
Consequently II A + B A II+ II B II. (d) I ABX '<II A II I B X 1 <II A 11 IIBII 1 X 1 . Applying this inequality to the vector X0, for which we get II AB ! Xol= ABXol
II AB II<II A II IIB II (e) Let A=p+qi be an eigenvalue of the matrix A, X= Y+iZ the corresponding eigenvector. Then
AY=pYqZ,AZ=qY+pZ
whence I Y12,
I qY+pZI 2 <IIA112 I Z12.

Combining these inequalities, we get 1, 12 y12+1 z 12) (p2 92) ( I
1 2 +1 Z 12)
A 112
Y [2+I. Z 12)
OH. 7. LINEAR ALGEBRA
305
and, hence,
IAIsII A II.
957. Let A be a real nonsingular matrix; then AAX X= I AX II is a positive quadratic form which can be brought to canonical form by a transformation of the variables with the triangular matrix B which has positive diagonal elements. Therefore, AA=BB, whence it follows that AB-' AB-'= =E, that is, AB-1.=P is an orthogonal matrix. From this we have A=PB. Uniqueness follows from the uniqueness of the triangular transformation of a quadratic form to canonical form. 958. AA is a symmetric matrix with positive eigenvalues a,, ..., An . Consequently,
AA = PWe construct the matrix
P.
B=P
where iis the positive square root of Al. It is obvious that B is again a symmetric matrix with positive eigenvalues and B2= AA -. Whence it follows that AB-1=Q is an orthogonal matrix, A= QB. 959. After carrying the coordinate origin to the centre of the surface, the surface must contain point - X along with point X, and, hence, the equation must not contain the running coordinates to first powers. After a parallel translation of the axes X= X0+ X', where X0 is the translation vector, the equation of the surface becomes
AX' X' +2 (AX0+B) X' + AX0 X0+2BXo+ C=0.

Therefore, for the existence of a centre, it is necessary and sufficient that the equation A X0+B be solvable with respect to the vector X0, for which, in turn, it is necessary and sufficient that the rank of the matrix A be equal to the rank of the matrix (A, B). 960. After carrying the origin to the centre, the equation of the surface becomes
AX X+y=0.
If r is the rank of the matrix A and al, oc, are nonzero eigenvalues, then after an appropriate orthogonal transformation of the coordinates, the equation takes on the canonical form oci4+ ... ar 4+ y=0. 961. The surface has no centre if the rank of (A,1 B) is greater than the rank of A, which is only possible if r= rank A <n. Denote the whole space by R, the space AR by P and the orthogonal complement of P by Q. Then for any YEQ we will have AY=0 because I AY 12= AY AY= Y AA Y=0
306
PART III
ANSWERS AND SOLUTIONS
B=B1+B2, Bi EP, B2EQ Then B2 00, otherwise B would belong to P and the rank of (A, B) would be equal to r. Let B2=AX,. After the translation X= X0 + X', the equation of the surface becomes AX' X' +2B2 X' + c' =0.
Make one more translation X' = aB2+ X". Then AX' X' =a2AB2 B2 +2aAB2 X" + AX" X"= AX" X" because AB2 =0 and, hence, the equation becomes AX" X"+2B2 X" +2a I B2 1 2 + c' =O. Take a-
since AA YEP. Let
c'
2 B, 12
Then the equation becomes AX" X"+2B2 X"=0. Now perform an orthogonal transformation of the coordinates, taking pairwise orthogonal unit eigenveaors of the matrix A for the basis of the space, including in them the unit vector that is collinear with the vector B2 and in the opposite direction. This can be done because B2is an eigenvector of A. On this basis, the equation becomes A2.7c;. + 23,x, +1 = where (32= I By I. It remains to divide by [3.9. 962. Under a linear transformation with the matrix A, the space is mapped onto the subspace spanned by the vectors A1, Ay, ..., Anwhose coordinates form the columns of A. The required result follows immediately. 963. Let e1, e2, eabe the basis of Q. Then Q' is a space spanned by 7 , where are the images of el, e2, e: under the linear transformation thus performed. Hence, Besides, it is obvious that q' because Q' lies in R'. Furthermore, let P' be the complementary space of Q, of dimension p= n q, and let p' be its image under a linear transformation. Its dimension p' does not exceed nq. But P'+ Q' =R', hence, p' +q' r. From this, 964. Let the rank of A be r1, the rank of B be r2 and let BR= Q. The dimensionality of Q is equal to r2. Then p, which is equal to the rank of AB, is the dimensionality of ABR= AQ. By virtue of the result of the preceding problem, ri+r2 n- p min (r9 , r2). 965. A double performance of the operation of projection is equivalent to a single performance. Indeed, in the first projection, all the vectors of the space R go into the vectors of the subspace P, which under the second projection remain fixed. Hence, A2=A. Conversely, let A2= A. Denote by P the set of all vectors Y=AX, by Q the set of vectors Z such that AZ=0. It is obvious that P and Q are linear spaces. Their intersection is the zero vector, because if AX=Z, then AX= A2X= AZ=0. Furthermore, for any vector X we have the expansion X=AX+(EA) X. It is obvious that (E A) XeQ because A (E A) X=(A A') X=0. Therefore, P + Q is the whole space, that is, P and Q are mutually complementary subspaces. The operation AX is
307
a transition from the vector X to its component in P, that is, it is the operation of projection on P parallel to Q. 966. Let P _LQ. Choose an orthonormal basis of the whole space by combining the orthonormal bases of P and Q. In this basis, the projection matrix will have the diagonal form 1 1
A'
1 0 0
In any other orthonormal basis, the projec ion matrix is equal to A= =B-1A'B, where B is some orthogonal matrix. Obviously, A is symmetric. Conversely, if A is a symmetric matrix and A =A, then the spaces P=AR and Q=(E A) R are orthogonal, because A X (E A) Y = X A (E A) Y = X (A A2) Y=0. 967. Let A be a skew-symmetric matrix. It is easy to see that AX X=0 for any real vector X, because AX X= X AX= X ( AX)= AX X. Let A= + Pi be an eigenvalue of the matrix A and U= X+ Yi its corresponding eigenvector. Then AX=ccX-13 Y, AY= px+. Y. From this it follows that a( 1 X12 +1 Y12 )= A X X+ AY Y=0 and m=0. Furthermore, px Y+ al Y12 = A Y Y=0, whence X Y=0 for (3o 0. Finally, from the equality P ( I X 1 2- I YI2)= AY X+ AX Y= AY X- X A Y=0 follows I XI=1 Yl 968. Let 0 a12 am A =( a 12 0 a2n

ain
al . . .
be a skew-symmetric matrix. If all its eigenvalues are equal to zero, then A=0. Indeed, the sum of the products of all eigenvalues taken two at a time is equal to the sum of all principal minors of the second order
i<k
a2k and the
fact that this sum is zero implies aik =0 for any i, k, that is, A=0. Let A have a nonzero eigenvalue A1= a1i. Normalize the real and imaginary parts of the eigenvector belonging to it. Because of the equality of their lengths, the normalizing factor will be the same, and the equations AX= ai Y, AY = a,X will hold for the resulting vectors.
308
Form an orthogonal matrix by putting the vectors X and Y in the first two columns. Then
0 I a, P-1AP-=0 \ 0
al 0 ... 0..
Since the matrix .1 1-"AP is skew-symmetric, all nonindicated elements of the first two rows are zero, and the matrix of the elements of the 3rd, 4th, ..., nth row and the 3rd, 4th, ..., nth column will be skew-symmetric. Argu0 a2 ing in similar fashion, we isolate yet another submatrix I. The pro02 0 cess is continued until the lower left corner has a matrix all eigenvalues of which are zero. But all the elements of such a matrix are zero. The problem is solved. 969. Let
-
B (E A) (E + A) - .
Then
B=(E+ A)-' (E A)=(E A) -' (E+ A)= B-'.

Furthermore
B+ E=(E A) (E+ A)-1+ (E+ A) (E+ A) -1= 2 (E+ A) -'

and, hence,
I B+ E 100.
Conversely, if
B=B-1and I B+ E 100 then for A we can take (E+ B)-' (E B). It is easy to s:e that A is a skewsymmetric matrix. 970. Let A be an orthogonal matrix. Then AX AY = X AAY= X Y
for any real vectors X and Y. Let X=cc-I-(3i be an eigenvalue of the matrix A, and
U= X+ Yi
a corresponding eigenvector. Then
AX=aXpY, A Y=oc Y+ f3X

whence
I X12= X X= AX AX=oc2I X 12+32 I Y 1 2 -24X Y, YI2 = Y- Y=A Y A Y=a2 I YI2 +132IX 12 +213X Y.
Adding these equations, we get 0+ p2= 1. 971. For SOO, we get, from the last equations of the preceding problem, 3 (1 X12-1 YI2 )+2ocX Y=0. On the other hand,
X Y= A X AY=(0 [32) X Y-E.p. (
x [2-
Y12)
7. LINEAR ALGEBRA
309
whence a (1 X12- 1 Yr)-2PX Y= 0. Therefore

X. Y =I X 121 Y12 =0.
972. 1. Let A= cc+ Pi= cos + i sin cp be a complex eigenvalue of the matrix. Form an orthogonal matrix A, the first two columns of which constitute the real and imaginary parts of an eigenvector belonging to A. Then coscp sin cp
Q=
sin cp cos cp 0 0 .. 0 0 0 0 ..
Because of the orthogonality of the matrix Q -',4Q, the sum of the squares of the elements of each row is equal to I, and, hence, all nonindicated elements of the first two rows are equal to 0. 2. Let A= +1 be a real eigenvalue of the matrix A, and let X be a normalized eigenvector belonging to A. Form an orthogonal matrix, the vector X constituting the first column. Then
11 . . .
AQ=(
0 ...
All nonindicated elements of the first row are zero since the sum of the squares of the elements of each row of the orthogonal matrix Q -1AQ is equal to unity. Applying the foregoing reasoning to the orthogonal matrices which remain in the lower left-hand corner, we get the required result. 973. (a) ( 1 0 0 0 2 0 0 0 0 0 1 0 1 0 0 0 0 1 0 , (k) , (b) ( 2 0 0 0' 1 1 , (h) (e) 2 0 0 1 0 1 2 0 0 1 0 0 1
2/
0 0 ), 1 , 1 0
(c) ( 3 0 0 , 0 0 1
(i) ( - 1
0 (1) 0 (2 0I0 0 00 (i) 1
(g) 2
0
(
0 2
1 0
0 0 0
0 1 0 0 0 0 1 0 \ 0
0
1 1 0
0l, 0 0/
0 0 0 0 1 0 0 0
0 i Q 0 0
(j)
1 ), 1/
310
3 0 0 0 2 0 1 (m)( 0 0 974. (a) 1 / 0

t
(n) ( 2 , 0 0
0 0 0 0 0 0
(o) ,
0 0 0
1 0 0
0 0 0)
1 1 1 0 1
(b) , 1 0
1
I 1 0
(c) 1 1 0 / 1 / 1
1 /
/I
/
'
27-c 27c where e=cos--- +1 sin
975. The submatrix
Ai Ai
1 1
cannot be periodic.
al1 /I
Remark. The result is not valid in a field with nonzero characteristic. For example, in a field of characteristic 2 we have
976. Let A be a given matrix and let B=C-'AC be its reduction to Jordan canonical form. The canonical matrix B is of triangular form and its diagonal elements are equal to the eigenvalues of the matrix A, each one being repeated as many times as its multiplicity in the characteristic equation. Furthermore, Bm ' c'n (Problem 531). Consequently, the characteristic polynomials of the matrices Am and Bn, coincide. Given an appropriate numbering of combinations, the matrix Bm ' has triangular form and, hence, its eigenvalues are equal to the diagonal elements. They are also, evidently, equal to all possible products of the eigenvalues of the matrix A taken m at a time. 977. The matrices A XE and AAE obviously have coincident elementary divisors. Therefore, A and A are convertible to cne and the same canonical matrix and, hence, they are interconvertible. 978. If A= CD, where C, D are symmetric matrices and the n-.atrix C is nonsingular, then A= DC and, hence, A= AC. Thus, the matrix C should be sought among matrices that transform A into A. Let A= SBS-', where B is a canonical matrix:
R. B=
By
where Bi=
'
Bk
141
Then A=-S 'BS. Denote by H, the matrix 1 1 1 1 It is easy to see that Bi=H, 1B1H1and, therefore,
111
B = H 1BH where H=
H2
Hk
Thus, A = S ' H ' BHS= S '11-1S-1ASHS=C-1AC where C= SHS . The matrix C is obviously symmetric. Put D= C IA. Then D= AC-1= C-1ACC-1= D. Thus, the matrix D is also symmetric and A= CD . 979. Let AAB' I =(_ i)n (X" c1 An 1_ c2 An 2 _ cn). It is obvious that pi=tr A =ci. Suppose we have already proved that Pi= =c1, p2=c2, , Pk I.= ck _i. Under this assumption, we shall prove that _ Pk_ iA=Ak pk = ck. By construction, Ak= Ak _piAk 1._p2Ak 2_ Hence ciA k C2Ak ... Ck_
tr A k= kp k= tr Ak citr Ak ck_1 tr A
= S k k_i C k
where S1, S2, Sk are p3xer sums of the eigenvalues of the matrix A. But C k_ i= 'xi,. Consequently, by Newton's formulas, S k CIS k_i
Pk = Ck
Next, Bn= An ciAn 1 . . . c_ 1A cE= 0 by the Hamilton-Cayley relation. Finally,
AB_ i= An= cE
whence
Bn_ i= cA -1.
980. Let 0 C= c21
c,,,
C12 0

Cln
Cyr,
Cn2 .. 0
We consider the diagonal matrix

a2
X=
0
312
PAR I III. ANSWERS AND SOLUTIONS
the diagonal elements of which are arbitrary but pairwise distinct. Let us take
bu
y=( bn ,
Then
0
b12 bin b22 1,2)
h n2 bnn
bi2 (a2 al) bin (an al)
X Y YX= (NI (al az) bm (al an)
...
b2n
(an 0
bn2 (c,2 an)

k
0 for ilk. Now, using
and, consequently, it suffices to take bik =
mathematical induction, we establish that any matrix with trace zero is similar to a matrix all diagonal elements of which are zero. Since tr C=0, CO ELE for E./.00, and, hence, there is a vector U such that CU and U are linearly independent. Including the vectors U and CU in the basis, we find that C is similar to the matrix
o o
112 Y22
Y13 Y23
Yin \ Yen Ynn

=
Y n2
Yn3
F )
It is obvious that tr P=tr C=0. Hence, by the induction hypothesis, r=s-i Ps where P' is a matrix with zero diagonal elements. Then all the diagonal elements are zero in the matrix C"=
PS
(0 1
SI ' C\ I0 1 S C)
(S-1-Q S-'FS)
INDEX
Adjoint matrices 85 Algebra higher, proof of fundamental theorem of 92ff linear 133ff, 166ff, 286ff of quaternions 208 Algorithm, Euclid's 97ff Alphabetical ordering 218 Approximation of roots of a poly nomial 115 Arithmetic progression 97 Base field 288 Basis (of a vector space) 133 orthonormal 136 Bezout's method 124 Biquadratic equation 96 Bounds of roots of a polynomial 107ff Bunyakovsky inequality 84, 215, 216 Cardan's formula 19 Cavalieri's principle 139, 293 Cayley (Hamilton-Cayley relation) 311 Characteristic polynomials 310 Chebyshev polynomial 128 Circular permutations 130ff, 285 Coefficients, method of undetermined 99 Cofactor 57 Complement, orthogonal 305 Complementary minors 85 Complex number(s) 1 I , 15 I ff, 168ff in trigonometric form 13 Condition, Sylvester's 142, 299 Coordinate space, n-dimensional 137 Coordinates, transformation of 133ff Cranier's theorem 61 criterion, Eisenstein 104, 239 Cubic equation 95, 118 Cyclic determinant 53, 56 Cyclotomic polynomial(s) 21, 95, 127 d 'Alembert's lemma 92, 161 Decomposition into partial fractions 102, 259 de Moivre's formula 151, 171 Denominator, rationalization of 129ff Determinant(s) basic properties of 29ff computation of 31ff cyclic 53, 56 definition of 27, 28 evaluation of 25ff, 153ff, 186ff expansion of 31 multiplication of 51ff order of 28 skew-cyclic 56 skew-symmetric 29, 56 Vandermonde 130, 165, 158, 164, 196, 222, 234, 284 Dimension (of a vector space) 133 Discriminant 19, 124ff Distance, definition of 137 Eigenvalues 139ff, 295ff, 304ff Eigenvectors 139ff Eisenstein criterion 104, 239 Eisenstein theorem 162, 239 Elementary symmetric polynomials 116ff, 265ff Equation(s) biquadratic 96 cubic 95, 118 generalized reciprocal 229 linear, systems of 61ff, 68, 196ff quadratic 121 quartic 95, 96 reciprocal, gepeTalized 229 sextic 97
314
INDEX
systems of linear 61ff, 68, 196ff transformation of 123ff Euclidean space 133 elementary geometry of n-dimensional 135ff Euclid's algorithm 97ff Euler formula 249, 250 Euler theorem 248 Euler's identity 214 Euler's resolvent 269 Evaluation of determinants 153ff, 186ff Factorization into irreducible factors 93ff into linear factors 93ff Ferrari's method (solving the guartic equation) 20, 106 Ferrari's resolvent 127, 269 Field, of reals 93 of residues modulo 288 Form(s) Jordan canonical 146ff, 310 Jordan normal 148 linear, systems of 66ff quadratic 14Iff trigonometric, complex numbers in 13 Formula(s) binomial, Newton's 151 Cardan's 19 de Moivre's 171 Euler 249, 250 Lagrange 100, 161, 162 Lagrange interpolation 101, 161 Newton's 121, 267-268, 311 Newton's binomial 151 Taylor's 88, 91, 224-225 Fractional rational functions 100ff Fractions, partial, decomposition into 102, 259 Function(s) fractional rational 100ff monogenic symmetric 118 of one variable 88ff, 160, 221ff Sturm 108 symmetric 116ff, 164ff, 261 ff Fundamental system of solutions 74, 75 Fundamental theorem of higher algebra, proof of 92ff
General solution of a system of equations 75 Generalized reciprocal equations 229 Hamilton-Cayley relation 311 Hermite method 124 Hermite polynomial(s) 110, 128, 267, 280 Higher algebra, proof of fundamental theorem of 92ff Homer's scheme 88, 89 Hypothesis, induction 220 Identity, Euler's 214 Induction hypothesis 220 Inequality, Bunyakovsky 84, 215, 216 Inertia, law of 300 Interpolation problem 100ff Inversions (in permutations) 26, 27 Irreducibility over the field of rationale 103ff Isomorphism 80, 207 Jordan canonical form 146ff, 310 Jordan normal form 148 Kronecker product 86, 141, 219 Lagrange formula 100, 161, 162 Lagrange interpolation formula 101, 161 Lagrange's method of multipliers 165 Laguerre polynomial 110, 128 Laplace theorem 49, 159, 215, 216 Law of inertia 300 Legendre polynomial 111 Lemma, d'Alembert's 92, 161 Linear algebra 133ff, 166ff, 286ff Linear equations, systems of 61ff, 68, 196ff Linear manifolds 133ff, 288 Linear transformations 146ff, 306 Linearly dependent solutions 74 Linearly independent solutions 74
INDEX
315
Manifold(s) linear 133ff, 288 one-dimensional 133 two-dimensional 133 Matrices (see matrix) 76ff, 159ff, 203ff adjoint 84 postmultiplication of 81 premultiplication of 81 rectangular 83ff symmetric 141ff unimodular 82, 211 Matrix (see matrices) bordered 79 norm of 145 projection 307 rank of 64, 65 skew-symmetric 307 triangular 86 Method Bezout's 124 Ferrari's 20, 106 Hermite's 124 Lagrange's (of multipliers) 165 Newton's 100, 101 of undetermined coefficients 99, 162, 165 Minors, complementary 85 Monogenic polynomials 116 Monogenic symmetric function 118 Multipliers, Lagrange's method of 165 n-dimensional parallelepiped 138 Newton's binomial formula 151 Newton's formulas 121, 267-268,311 Newton's method 100, 161 Norm of a matrix 145 Number(s) complex 11, 151, 168ff in trigonometric form 13 One-dimensional manifolds 133 One-dimensional parallelepipeds 293 Ordering, alphabetical 218 Orthogonal complement 305 Orthogonal projection 147 Orthogonalization process 136 Orthonormal basis 136
Parallelepiped, n-dimensional 139 one-dimensional 293 Partial fractions, decomposition into 102 Permutations 26, 27 circular 130ff, 285 Polygon, regular I4-sided 22 Polynomial(s) bounds of roots 107ff characteristic 310 Chebyshev 128 cyclotomic 21, 95, 127 elementary symmetric 116ff, 265ff Hermite 110, 128, 267, 280 Laguerre 110, 128 Legendre Ill monogenic 116 operations on 88ff rational roots of 103ff roots of, approximation of 115ff roots of, distribution of, theorems on Illff Sturm 108, 246ff symmetric 119 unchanged under circular permutations of variables 130ff unchanged under even permutations of variables 130ff Postmultiplication of matrices 81, 209 Power sums 121, 143 Premultiplications of matrices 81, 208-209 Primitive roots 178ff, 183ff Primitive roots of unity 21-24 Principle, Cavalieri's 139, 293 Projection 147 operation of 306 orthogonal 147 Projection matrix 307 Product, Kronecker 86, 141, 219 Progression, arithmetic 97 Quadratic equation 121 Quadratic for ms 141ff Quartic equation 96 Quaternions, algebra of 208 Rank (of a matrix) 64-65 Rational roots of polynomials 103ff Rationalization of denominator 129ff
316
INDEX
Reals, field of 93 Reciprocal equations, generalized 229 Rectangular matrices 83ff Recurrence relations 162 Reducibility over the field of rationale 103ff Regular simplex 136 Relation, Hamilton-Cayley 311 Relations, recurrence 162 Resolvent Euler's 269 Ferrari's 127, 269 Resultant 124ff Rolle's theorem 162 ,251 ,253 Roots bounds of (of a polynomial) I07ff of a polynomial approximation of 115ff distribution of, theorems on I Ilff primitive I 78ff, 183ff rational (of polynomials) 103ff of unity 21ff primitive 21-24 Scheme, Horner's 88, 89 Sequence, Sturm 109, 110, 246 Sextic equation 97 Similarity transformation 293 Simplex, regular 136 Skew-cyclic determinant 56 Skew-symmetric determinant 29, 56 Skew-symmetric matrix 307 Solution(s) fundamental system of 74, 75 general (of a system of equations) 75 linearly dependent 74 linearly independent 74 system of, fundamental 74, 75 Space coordinate, n-dimensional 137 Euclidean 133 elementary geometry of ndimensional I 35ff vector 133 Sturm functions 109 Sturm polynomials 108, 246ff Sturm sequence 108, 109, 246ff Sturm theorem 108ff, 248 Submatrices 86, 87 Subspaces 133ff Sums, power 121ff, 143
Sylvester's condition 142, 299 Symmetric functions 116ff, 164ff 261ff Symmetric matrices 141ff Symmetric polynomial 119 System(s) of equations, general solution of 75 of linear equations 61ff, 68, 196ff of linear forms 66ff of solutions, fundamental 74, 75 Taylor's formula 88, 91, 224-225 Theorem(s) Cramer's 61 on distribution of roots of a polynomial Illff Eisenstein 162, 238 Euler 248 fundamental (of higher algebra) proof of 92ff Laplace 49, 159, 215, 216 Rolle's 162, 251, 253 Sturm's 108, 247 tr (see trace) Trace of a matrix (tr) 150 Transformation of coordinates 133ff of equations 123ff linear I46ff, 306 similarity 293 triangular 142 Tschirnhausen 129ff, 284 Transpose (of a matrix) 83 Triangular matrix 86 Triangular transformation 142 Trigonometric form, complex numbers in 13 Tschirnhausen transformation 129ff, 284 Two-dimensional manifolds 133 Undetermined coefficients, method of 99, 162, 165 Unimodular matrices 82, 211 Unity, roots of 21ff Vandermonde determinant 130, 156, 158, 164ff, 196, 222, 234, 284 Vector space 133 Vectors 133

MIR - Faddeev D. K. and Sominsky I. - Problems in Higher Algebra - 1972

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

MIR - Faddeev D. K. and Sominsky I. - Problems in Higher Algebra - 1972

Загружено:

Авторское право:

Доступные форматы

A. K. (DALLEIEEB, H. C.

PROBLEMS IN HIGHER ALGEBRA

MIR PUBLISHERS MOSCOW

UDC 512.8 (075.8)=20

Revised from the 1968 Russian edition

Printed in the Union of Soviet Socialist Republics

Contents Introduction Part I PROBLEMS

I. Operations on Square Matrices 2. Rectangular Matrices. Some Inequalities

PART II HINTS TO SOLUTIONS

151 153 159 160 164 166

7 PART III ANSWERS AND SOLUTIONS

168 186 196 203 221 261 286 313

CHAPTER 1 COMPLEX NUMBERS

Sec. 1. Operations on Complex Numbers

3. Evaluate in, where n is an integer. 4. Verify the identity x4 +4=(x-1-0(xl+i)(x+1+i)(x+1-0.

(d) (1 (1+05 +1 ' (e) (1 + 0n y 8. Evaluate (1 i

where n is a positive integer.

9. Solve the following systems of equations:

CH. 1. COMPLEX NUMBERS

i1/2, 1+i, 2-3i.

(f) 1+ i, (g) 1 i, (h) 1 i (i) 1 + i V 3 ,

x x=1 +2i, (b)

*27. Prove the identity

What geometrical meaning does it have?

32. Demonstrate that if z

(b) (1 +1/ -v 3)2o, 1i \24

(d) (-1+ii/ 3)" (-1(1 _F i1 iro /

*37. Prove that

(a) (1 + = 2 (cos 4- + i sin 2'4`) 7 ,

CH. I. COMPLEX NUMBERS

42. Prove that

( 1 + itan oc )22 i tan a

1 + i tan na 1 i tan not

43. Extract the roots:

(b) 1/2 2i, (c) 1/ 4, (d)

44. Use tables to extract the following roots :

*51. Prove that m-1 (a) 22m cos2m x = 2

(b) 22m cos2m +1 x=

CL +1cos (2m 2k +1) X

(c) 2'm sin2m x = 2

(d) 2'm sin2m+1 x=E (_,),n+k Ct+Isin (2m 2k + 1) x.

*52. Prove that 2 cos mx = (2 cos x)m

x (2 cos x)in -2P

(a) 1 0+0,-0+ ..., (b) ++...

+2 2COS 17r 4)'

'4' - . Cn " +... = 2 1 (2n -' 2 2 sin 17

CH. 1. COMPLEX NUMBERS

*56. Find the sum

(b) Gin + C,1+

(24+ 2 cos (n -3 ) (2" + 2 cos (n-3 )7

59. Compute the sums :

sin x + sin 2 x + . . . + sin nx = 61. Find

1 1 , 1 ( 1 + cos x + 71cos 2x + . .. + yn- cos nx) .

63. Show that

+ sin2 (2n 1)x.

2 sin x cos (n + 1)x sin nx 2 sin x

CH. I. COMPLEX NUMBERS

Sec. 3. Equations of Third and Fourth Degree

(o) x4 2X3 X2 +2x-1=0,

CH. 1. COMPLEX NUMBERS

*87. Let e be a primitive 2n-th root of unity. Compute the sum

2Tc Tr /Mt (b) sin 2m+ 1 sin 2m+ 1 . . . sin 2m +1

*100. Prove that fl(a + bek )= an + (

*101. Prove that